neither intent nor impact: a critique of the racially

Neither Intent nor Impact:

A Critique of the Racially Based

Selective Prosecution Jurisprudence

and a Reform Proposal

Yoav Sapir∗

The doctrine controlling equal protection challenges of racially basedselective prosecution is far from being satisfactory. It is hardly ever thecase that a defendant can prove that a prosecutor had an invidious motivein her prosecution. Even if we generally accept the discriminatory intentas the heart of equal protection challenges, we should reject it in the speciªccontext of racially based selective prosecutions. Special features of crimi-nal law, of race issues, of prosecutorial discretion, and especially the com-bination of the three, mandate a new standard that forgoes the intent re-quirement. Yet the alternative that is usually presented—a standardbased on proof of disparate impact—is no less problematic, because it mightrender the criminal justice system dysfunctional. This Article suggests anew double-stage standard that will transcend the pitfalls of the intentrequirement without running the risks of a pure impact standard. In theªrst stage, a defendant, who claims that she was singled out for prosecu-tion on the basis of her race, will have to show that a similarly situated indi-vidual from a different race was not prosecuted. In the second stage, theprosecution will have the opportunity to prove that there is no correlationbetween race and the decision to prosecute by presenting statistical evi-dence. If applied, this standard will best serve the idea of equal protectionunderstood as the prohibition of unequal treatment.

The prosecutor has more control over life, liberty, and reputa-tion than any other person in America. —Attorney General, Robert Jackson, 19401

∗ S.J.D. candidate, Clarke Byse Fellow, and Teaching Fellow, Harvard Law School;

LL.M., Harvard, 2000; LL.B., Tel-Aviv University, 1996. I am grateful to Carol Steikerfor her continuous support and for her criticism and guidance. I would also like tothank Yishai Blank for his comments on earlier drafts, and the editors of the HarvardBlackLetter Law Journal, especially Ricardo Exantus and Bridgette Thornton, for theirimportant substantive as well as technical work.

1. Attorney General Robert Jackson, The Federal Prosecutor, Address Delivered at theSecond Annual Conference of United States Attorneys (Apr. 1, 1940), 24 J. Am. Judi-

cature Soc’y 18, 19 (1940).

128 � Harvard BlackLetter Law Journal � Vol. 19, 2003

The prejudices of centuries die hard, and even when they wane,the institutional frameworks that sustained them are bound to lin-ger. —Orlando Patterson, 19972

I. Introduction

It is not new that race and ethnicity play a major role in law enforce-ment. The unequal treatment of people of Middle-Eastern descent in thewake of September 11 attacks3 and the commutation of all death sentencesby Illinois Governor Ryan4 are just current reminders of this disturbingphenomenon. Does the legal system try hard enough to ameliorate thisproblem? I argue that, in the ªeld of prosecutorial discretion, it does not,although it can. The purpose of this Article is to suggest an alternativestandard to the existing one for examining equal protection challenges ofselective prosecution, in which the defendant argues that she was singledout for prosecution on the basis of her race or ethnic origin.

The existing jurisprudence of equal protection challenges is held cap-tive in a discourse of intent versus impact, in which intent prevails. Myprimary assertion is that the current test that is used by the courts, ac-cording to which the defendant has to prove intent to discriminateagainst her on the basis of her race or ethnic origin, is far from being satis-factory. Yet, other suggestions to adopt a test that will focus on disparateimpact have proven to be no less problematic. Thus, it seems that the ju-risprudence of equal protection in criminal prosecutions has reached adeadlock. On one hand, the requirement to prove discriminatory intenthas rendered equal protection challenges almost impossible. It is hardlyever the case that a defendant can prove that a prosecutor had an actualintent to discriminate against her on the basis of her race. On the otherhand, it seems that the Court fears that waiving the intent requirementand being satisªed by a showing of disproportionate impact will paralyzethe criminal justice system altogether, since almost every rule has somesort of disparate impact. In other words, while intent is nowhere, impactis everywhere.

This deadlock could be witnessed in the results of recent SupremeCourt decisions. In the past seven years two cases that raised the issue ofracially based selective prosecution came up to the Supreme Court. Each

2. Orlando Patterson, The Ordeal of Integration: Progress and Resentment in

America’s “Racial” Crisis 15 (1997).3. For a detailed description of the unequal treatment of mostly (but not only) immi-

grants on the basis of their national origin by law enforcement authorities in thewake of September 11, and for a compelling critique of that response, see David Cole,Enemy Aliens, 54 Stan. L. Rev. 953 (2002). For the role of racism in that response seeMari Matsuda, A Dangerous Place: A Response to David Cole’s “Their Liberties, Our Secu-rity,” B. Rev., Dec. 2002-Jan. 2003, at 20. For its relation to racial proªling, see SamuelR. Gross & Debra Livingston, Racial Proªling Under Attack, 102 Colum. L. Rev. 1413(2002).

4. The death penalty was handed out differently, Ryan said, depending on where peo-ple lived in Illinois, who their prosecutor was, who their defense lawyer was, howpoor they were and what race they were. See Maurice Possley & Steve Mills, Clem-ency for All, Chi. Trib., Jan. 12, 2003 at C1.

Racially Based Selective Prosecution Jurisprudence � 129

of these cases involved a separate area of criminal justice frequently chargedof being implicated with racial biases—the ªrst being the “war on drugs”and speciªcally the prosecution of crack-cocaine offenders5 and the sec-ond being the administration of the death penalty.6 However, the defen-dants in both cases did not even prevail in their motions for discoverythat would have possibly enabled them to prove a stronger case of dis-crimination. In both cases the Supreme Court reversed the decisions ofthe courts of appeals, which granted the discovery motions.

I will suggest a way out of this impasse by questioning the desirabil-ity of the intent doctrine in the area of selective prosecution and by pro-posing an alternative standard. Although there is a massive body ofscholarship on equal protection, on race issues, and on selective prosecu-tion claims, there has not been much criticism of the desirability of theintent doctrine in the speciªc ªeld of racially based selective prosecution.I will point out the uniqueness of this ªeld and the need to isolate it fromother equal protection issues. Thus, I will bracket the general issue of in-tent versus impact in equal protection claims in favor of a concrete scru-tiny of the intent test as it is applied to a speciªc area. Finally, I will intro-duce a proposal for a new double-stage test—similarly situated individu-als and comprehensive statistics—and argue that this proposal overcomesthe deªciencies of the intent requirement, while avoiding the risks andpitfalls of a “pure” disparate impact test.

A. Preliminary Clariªcations

In this Article, I consider the problem of unequal treatment from thedefendants’ perspective—that is, when defendants claim they were sin-gled out for prosecution on the basis of their race. Another problem in thecriminal justice system is discrimination based upon the race of the vic-tim. This phenomenon raises different questions that are beyond thescope of this Article. In addition, since my focus is on the criminal defenseof “selective prosecution,” I will not address here the possible remediesachievable through other proceedings like civil rights actions.7 I mainlydiscuss one form of unequal treatment in prosecutorial discretion, in whichcharges are brought against some people but not against others. I do notdiscuss other forms of discrimination, such as unequal treatment in plea-bargaining and in exercising the prosecutorial discretion whether to seekthe death penalty, whether to grant immunity, what charges to bring, orwhether to prosecute juveniles as adults, as well as whether to prosecutesome people in federal courts and others in state courts. Nonetheless, Ibelieve that, although these other modes of discrimination raise some ad-ditional questions, the model I develop in this Article can form the basisfor similar, adjusted models applicable to these other modes of unequaltreatment.

The most infamous unequal treatment by the criminal justice systemof the United States is the discrimination against African American peo-

5. United States v. Armstrong, 517 U.S. 456 (1996).6. United States v. Bass, 536 U.S. 862 (2002).7. Angela Davis is addressing both these issues in Prosecution and Race: The Power and

Privilege of Discretion, 67 Fordham L. Rev. 13 (1998).


ple compared to white people. For this reason, I often use the words “black”and “white” in this Article. However, I certainly mean for the discussionto apply to discrimination against other racial or ethnic minorities as well.

A ªnal clariªcation regards the scope and aim of this Article. Whenwe look at the “bottom line” of the criminal justice system, we see thatmore than half of the U.S. prison population is black, though blacks makeup only about thirteen percent of the general U.S. population. This Articleis certainly motivated by these alarming statistics, but it does not pretendto propose a solution that will solve the problem of race in the criminaljustice system. Race and law enforcement are completely intertwined. It isdifªcult to tell at what stage race is most signiªcant or inºuential. Is it ra-cial proªling by the police? Is it prosecutorial discretion? Jury selection?Sentencing? My proposal is only intended to suggest a reform that willreduce the phenomena of unequal treatment by focusing on the chargingdecision. It does not pretend to remedy (though it may to some extent)many law enforcement costs inºicted disproportionately on minorities,such as stops, frisks, searches, arrests, and police brutality, as well as thedisproportionate impact of substantive criminal law.

B. Race and Prosecutorial Discretion: Empirical Evidence of the Problem

Bearing in mind the other forms of discrimination in the criminal jus-tice system, is prosecutorial discretion an area worth focusing on? In PartII, I point out the uniqueness of racially based selective prosecution, and Iprovide explanations for the need to relinquish the discriminatory intentrequirement in this ªeld and to adopt an alternative standard that is moresensitive to the problem of unequal treatment. One of my arguments isthat discrimination and unequal treatment are at least as likely to occur inprosecutorial discretion as in other stages of the criminal process. Here, Isupport the theoretical argument by presenting several examples of em-pirical studies supporting the claim that there is a real problem of racialbias in the speciªc ªeld of prosecutorial discretion in the United States.8

The studies suggest that racial bias is present in every stage of prosecuto-rial discretion, including the decision to investigate, the initial assessmentof the severity of the offense, the charging decision, and the decisionwhether to seek the death penalty in homicide cases.

A few studies examined racial bias in prosecutorial discretion inFlorida homicide cases. William Bowers and Glenn Pierce conducted astudy of indictments for ªrst-degree murders between 1972 and 1977 inselected Florida counties. According to the study, the probability of beingindicted for ªrst degree murder for a black person who kills a white per-son was 0.925, while the probability of being indicted for ªrst degreemurder for a white person killing a black person was less than half—only0.429. Furthermore, homicide cases characterized by the police as in-volving no felony circumstances, or only suspected felony circumstances,were most likely to be upgraded by the prosecutor to felony murder if thedefendant was black and the victim was white. In cases in which the po-

8. For reasons of brevity, I mainly present the conclusions of the different studies and

not their methodologies or the different variables they control for. The interestedreader can refer to the actual sources cited below.


lice reports indicated felony circumstances and the victim was white,white defendants had a much better chance than did black defendants tohave their charge downgraded by the prosecutor. This selectivity, ac-cording to Bowers and Pierce, was the key reason for the high proportionof blacks on death row who had killed whites. In addition, the authorsconcluded that in light of the data on relative likelihood of convictions,there is an indication that blacks who killed whites were overcharged.9

Another study by Bowers, Pierce, and John McDevitt examined mur-der indictments in twenty Florida counties between 1976 and 1977. Thedata suggested that racial considerations lead prosecutors to “upgrade”some cases by alleging aggravating felony circumstances or charging thedefendant with an accompanying felony, and to “downgrade” others byignoring evidence in police reports or withholding an accompanyingcharge, depending on the race of the offender and the victim. The authorsfurther conducted a multiple regression analysis to determine how theindictment decisions in ªrst-degree murder cases were inºuenced by ex-tralegal factors. The data indicated that the race of both the defendant andthe victim affected the probability that the prosecutor would obtain aªrst-degree murder indictment. The authors concluded that, since the de-cision to seek a ªrst-degree murder indictment is almost exclusively un-der the discretion of the prosecution, the data indicated racial discrimina-tion on the part of prosecutors.10

Finally, in a comprehensive study examining 1017 homicide defen-dants in Florida, Michael Redelet and Pierce reached similar conclusions.The study focused on cases that differ in their police and prosecutorialclassiªcations. According to the ªndings both the race of the defendantand the race of the victim affected prosecutorial discretion. It was found,for example, that in cases in which the victim was white and the policeclassiªed the circumstances as non-felony, 63.6% of the cases remainedwithout the suggestion of an accompanying felony in the court data whenthe defendants were blacks, but 93.3% of the cases remained without suchsuggestion when the defendants were whites. After controlling for theinitial police determination, cases involving blacks accused of killingwhites were found to be the most likely to be “upgraded,” and cases in-volving whites accused of killing blacks were the least likely to be up-graded. The ªnal conclusion following a multivariate analysis controllingfor eight other possibly inºuential factors was that race, in effect, func-tions as an implicit aggravating factor in homicide cases.11

Race was also found to be inºuential on prosecutorial discretion inother places in the country and in other types of offences. In 1980, GaryLafree published the results of a study that examined screening decisionsand charging decisions in 881 sexual assault cases in a large midwesterncity. The study showed that black men who assaulted white women weremore likely to have their cases ªled as felonies and to receive more seri-

9. William Bowers & Glenn Pierce, Arbitrariness and Discrimination under Post-Furman

Capital Statutes, 26 Crime & Delinq. 563, 611–14 (1980).10. William J. Bowers, Glenn L. Pierce & John F. Mcdevitt, Legal Homicide: Death

as Punishment in America, 1864–1982, 340–44 (1984).11. Michael Radelet & Glenn Pierce, Race and Prosecutorial Discretion in Homicide Cases, 19

Law & Soc’y Rev. 587, 601–05, 615 (1985).


ous charges. Thus, for example, black men accused of assaulting whitewomen accounted for 23% of all reported rapes, but accounted for 45% ofall men sent to a state penitentiary and for 50% of those who were im-prisoned for six years or more. LaFree concluded that racial compositioninºuenced both the screening and the determination of the seriousness ofthe charge.12

A study performed by Cassia Spohn, John Gruhl, and Susan Welch ex-amined the prosecutor’s initial decision whether to prosecute and the de-cision to dismiss a felony charge after formally ªling it. The data revealedthat, in Los Angeles, while 59% of whites had their charges rejected, only40% of blacks and 37% of Hispanics had their charges rejected. Theauthors concluded that it “is also apparent that Hispanics are prosecutedmore often than blacks who are prosecuted more often than Anglos.”13

Another Los Angeles study examined data regarding the decision whetherto prosecute crack cocaine defendants in state courts or in the federal courts,in which they face much harsher mandatory minimums. It was foundthat, in state prosecutions, there were 222 whites and 4410 blacks. In fed-eral prosecutions, there was not even a single white defendant, but therewere 36 black defendants and 7 defendants from other minority groups.14

Racial bias in prosecutorial discretion crosses the lines of class, and af-fects not only the poor, but also the prominent and the powerful. A 1990study showed that fourteen percent of public corruption investigations con-ducted in the past ªve years was against black public ofªcials, whereasonly two percent of national elected ofªcials are blacks. In the South, thenumbers were even more striking: forty percent of the investigations wereagainst black public ofªcials, while only three percent of the elected pub-lic ofªcials were blacks. Out of ªve United States District Court judgesindicted within the last decade, three were blacks, although the overalloccupation of district judicial position by blacks is only one half of onepercent. Julius Chambers, the director of the NAACP Legal Defense Fundin New York City, said that “black leaders are being investigated morefrequently than their white counterparts simply because of their race.”15

Finally, in a recent case, the Court of Appeals for the Sixth Circuitafªrmed the ruling of the district court—that stark discriminatory effectof federal death penalty protocol, when coupled with ofªcial statementsof members of the Department of Justice, constituted at least some evi-dence that tended to show that race played a role in deciding what de-fendants to charge with death-eligible offences. The respondent, a blackperson who claimed that the Government had determined to seek thedeath penalty against him because of his race, has presented statistical

12. Gary LaFree, The Effect of Sexual Stratiªcation by Race on Ofªcial Reactions to Rape, 45

Am. Soc. Rev. 842 (1980).13. Cassia Sphon, John Gruhl & Susan Welch, The Impact of the Ethnicity and Gender of

Defendants on the Decision to Reject or Dismiss Felony Charges, 25 Criminology 175, 184(1987).

14. Richard Berk & Alec Campbell, Preliminary Data on Race Crack Charging Practices inLos Angeles, 6 Fed. Sentencing Rep. 37 (1993).

15. Julius Chambers quoted in Mark Curriden, Selective Prosecution—Are Black OfªcialsInvestigative Targets?, A.B.A. J. Feb. 1992, at 54, 55.


evidence, relying heavily on a 2000 Department of Justice report.16 TheCourt of Appeals found that

[T]he evidence shows that although whites make up the majorityof all federal prisoners, they are only one-ªfth of those charged bythe United States with death-eligible offenses. The United Statescharges blacks with a death-eligible offense more than twice asoften as it charges whites . . . . In addition, the United Statescharges blacks with racketeering murder one-and-a-half times asoften as it charges whites, and with ªrearms murder (Bass’scharge) more than twice as often as it charges blacks. Amongdeath penalty defendants, the United States enters plea bargainswith whites almost twice as often as it does with blacks.17

C. Washington v. Davis and Its Critiques

In Washington v. Davis,18 black applicants challenged the constitution-ality of an admission test to a police ofªcers’ training program. The fail-ure rate for blacks in this test was four times as high as for whites. In re-jecting the equal protection claim of racial discrimination, the SupremeCourt held that a law is not unconstitutional simply because it has a ra-cially disproportionate impact. Rather, in order for such a claim to be suc-cessful, one must show that the law reºects a racially discriminatory pur-pose. Thus, Davis, which is by now a well-established precedent, not onlysustained the distinction between intent and impact, but also declared thedefeat of impact as the standard for violation of the Equal Protection Clause.

The intent requirement of Davis was criticized on various grounds.Most importantly, it was argued that the intent doctrine failed to capturesocial inequalities that exist regardless of intentions;19 that it ignored the

16. U.S. Dep’t of Justice, The Fed. Death Penalty System: A Statistical Survey

(1988–2000) (Sept. 12, 2000), available at http://www.usdoj.gov/dag/pubdoc/dpsurvey.htm. The Department of Justice denies the claim that the data indicatesracial bias. See U.S. Dep’t of Justice, The Fed. Death Penalty System: Supplementary

Data, Analysis and Revised Protocols for Capital Case Review (June 6, 2001),available at http://www.usdoj.gov/dag/ pubdoc/deathpenaltystudy.htm.

17. United States v. Bass, 266 F.3d 532, 538 (6th Cir. 2001). According to the Departmentof Justice survey, ªrearms murder, racketeering murder, and continuing criminal en-terprise murder (the three charges brought most frequently against death-eligibleblacks) “can be charged in a wide array of circumstances, and [are] therefore morelikely to be available as a charging option in a given case than more narrowly deªned of-fenses such as kidnapping-related murder.” Supra note 16. The Court of Appeals’ruling to afªrm the ruling of the District Court, granting the discovery request of thedefendant, was reversed by the Supreme Court. Regarding the statistical evidence,the Court said: “Even assuming that the Armstrong requirement can be satisªed by anationwide showing (as opposed to a showing regarding the record of the decision-makers in respondent’s case), raw statistics regarding overall charges say nothingabout charges brought against similarly situated defendants. And the statistics regard-ing plea bargains are even less relevant, since respondent was offered a plea bargainbut declined it.” Bass, supra note 6.

18. 426 U.S. 229 (1976).19. See Theodore Eisenberg, Disproportionate Impact and Illicit Motive: Theories of Constitu-

tional Adjudication, 52 N.Y.U. L. Rev. 36 (1977). See also Alan David Freeman, Legiti-mizing Racial Discrimination Through Antidiscrimination Law: A Critical Review of the Su-


existence of unconscious racism in American society;20 and that even incases where there is discriminatory intent, it is impossible to prove it.21

D. The Application of Davis to Selective Prosecution

Despite the critiques, Davis’s rejection of the disparate impact test infavor of a discriminatory intent test permeated almost every particularbranch of equal protection doctrine. Thus, Davis was adopted also in crimi-nal cases with regard to both racially based and non-racially based selec-tive prosecution claims.

In Wayte v. United States,22 the claim was against the Government’s“passive enforcement” policy, according to which only those who reportedthemselves or were reported by others to have violated the law requiringdraft registration were prosecuted. Only thirteen out of an estimated 674,000people who failed to register for the draft were prosecuted. The petitionerargued that the impact of that policy was that only “vocal” opponents ofthe law were prosecuted, and that therefore it was a violation of the Firstand the Fifth Amendments. The Supreme Court held that “it is appropri-ate to judge selective prosecution claims according to ordinary equal protec-tion standards.”23 The “ordinary equal protection standard” was, natu-rally, the Davis standard. The Court explained that this standard requiredthe petitioner to show not only that the passive enforcement policy had adiscriminatory effect, but also that it was motivated by a discriminatorypurpose. The petitioner’s claim was dismissed because he failed to showthat the Government prosecuted him because of his protest activity. TheCourt noted that discriminatory intent “implies more than . . . intent asawareness of consequences. It implies that the decision maker . . . selectedor reafªrmed a particular course of action at least in part ‘because of,’ notmerely ‘in spite of,’ its adverse effects upon an identiªable group.”24

In United States v. Armstrong,25 the respondents claimed that they weresingled out for prosecution because of their race. In this case, the Courtconsidered the question of what proof was necessary for a defendant toshow in order to be entitled to discovery on such a claim. Although notthe main issue to determine in this case, the Court said that the require-ments for a selective prosecution claim draw on an “ordinary equal pro-tection standard,” and that the claimants had to show both that the prose-cutorial policy had a discriminatory effect and that this policy was moti-vated by a discriminatory purpose.26

preme Court Doctrine, 62 Minn. L. Rev. 1049, 1052–57 (1978) (Freeman argued that theintent requirement focuses on the “perpetuator perspective” which faultily sees dis-crimination as wrong conduct of one actor rather than as a social phenomenon).

20. Charles R. Lawrence III, The Id, the Ego, and Equal Protection: Reckoning with Uncon-scious Racism, 39 Stan. L. Rev 317 (1987).

21. See, e.g., Kenneth Karst, The Costs of Motive-Centered Inquiry, 15 San Diego L. Rev.

1163 (1978).

22. 470 U.S. 598 (1985).23. Id. at 608.24. Id. at. 610.25. Supra note 5.26. Id. at 465. On the discovery issue, the court considered what evidence constituted

“some evidence tending to show the existence of the discriminatory effect element,”and held that the defendant must produce credible evidence that passes a threshold


Wayte and Armstrong thus applied the intent requirement to equal pro-tection challenges of selective prosecution. Although numerous critiquesof the general intent requirement followed Davis during the late 1970sand early 1980s, hardly any academic work was done on the applicationof this doctrine to the ªeld of selective prosecution. One gets the feelingthat a frustration has developed among the scholars of the 1990s fromtrying to attack the well-entrenched status of the intent doctrine in equalprotection jurisprudence. Instead, writers focused on other aspects of theselective prosecution doctrine, such as remedies, procedures, and discov-ery. Their assumption was probably either that everything has alreadybeen said about the intent versus impact dilemma, or that there was sim-ply no chance to convince the Court (at least not the current one) tooverturn Davis and its applications.27 While the theoretical critiques of thegeneral intent doctrine rarely referred to the speciªc problem of selectiveprosecution and race, the current writings on selective prosecution andrace seldom discuss the desirability of the intent requirement, and theyseem to assume it to be unquestionable. This Article addresses this voidby arguing that regardless of our position on general equal protectionchallenges—intent, impact, or otherwise—with regard to racially based se-lective prosecution cases, the doctrine of discriminatory intent should bediscarded in favor of a test that is based on neither intent nor impact.

showing that similarly situated individuals of a different race could have been prose-cuted but were not. Id. at 469. Although it is somewhat ambiguous from the plainlanguage of Armstrong, in a recent decision the Supreme Court referred to Armstrongas holding that defendant must also show some evidence of discriminatory intent.The Supreme Court thus applied both the intent requirement and the effect require-ment to the discovery stage. See Bass, supra note 6; see also United States v. Jones, 159F.3d 969, 978 (6th Cir. 1998) (applying this standard and concluding that appellant isentitled to discovery. The intent prong was satisªed by proof of the arresting policeofªcers’ behavior in taunting defendant by the mailing of a racially charged post-card, and the wearing of custom-made T-shirts with inappropriate personalized lan-guage and pictures of defendant and his wife. Obviously, this kind of behavior isvery rare.); U.S v. James, 257 F.3d 1173 (10th Cir. 2001) (denying the motion for dis-covery on the basis of inability to show similarly situated individuals who were notprosecuted).

27. A few examples might be useful. Pamela Karlan, for instance, discusses selectiveprosecution and Armstrong, but focuses on the remedy question. The need to showdiscriminatory intent is neither questioned nor criticized. See Pamela S. Karlan, Race,Rights, and Remedies in Criminal Adjudication, 96 Mich. L. Rev. 2001, 2003 (1998).

Similarly, Andrew Leopold brieºy summarizes some of the critiques of the intent re-quirements, and then proceeds: “Whether we should require proof of a discriminatoryintent is an important, but at the moment, not a very practical question.” Andrew D.Leipold, Objective Tests and Subjective Bias: Some Problems of Discriminatory Intent in theCriminal Law, 73 Chi.-Kent L. Rev. 559, 563 (1998) (emphasis added). Instead, hechooses to focus on pretrial and trial practices and the ways in which they preventdefendants from raising discriminatory prosecution claims. Id. In an article that is acritique of Armstrong, Richard McAdams assumes “the correctness of basic equalprotection doctrine,” and forgoes “relying on criticism of Supreme Court precedentrequiring, as an element of equal protection claims, proof that governmental ofªcialsact with a discriminatory purpose.” Richard H. McAdams, Race and Selective Prosecu-tion: Discovering the Pitfalls of Armstrong, 73 Chi.-Kent L. Rev. 605, 606 n.7 (1998).McAdams concentrates only on the discovery issue of Armstrong. Id.


II. The Uniqueness of Racially Based Selective Prosecution

Although I agree with many of the general critiques of Davis, this Ar-ticle is limited to a suggestion to change the doctrine with regard to ra-cially based selective prosecutions. I will therefore show that, even if weassume Davis to be justiªed as the general equal protection doctrine, it isnot desirable as the standard for racially based challenges of selectiveprosecution. I will ªrst challenge the position that “coherence” requiresuniªed equal protection jurisprudence throughout the different ªelds oflaw. Next, I will show that the subject of this Article—racially based se-lective prosecution—is different than other equal protection challenges,and argue that these differences demand the rejection of the intent require-ment. The uniqueness of this topic is threefold: it involves criminal law,prosecutorial discretion, and race.

A. Coherence and Universalism in Equal Protection Law

The Court’s adoption of the “ordinary equal protection standards” inWayte and in Armstrong seems to be based on the assumption that a uniªedtest for all equal protection challenges is desirable. I would like to call thisassumption into question, and suggest an approach that takes into ac-count the speciªcity of different ªelds of law.

In his article on sentencing in crack-cocaine cases,28 David Sklanskyopposes the “universalist approach” of the Court. He mentions that thisapproach started even earlier than the intent requirement of Davis andthat the Court committed itself to promulgating a single, globally appli-cable set of equal protection rules as early as 1970 in Dandridge v. Wil-liams.29 Sklansky criticizes the aspiration for a uniªed test, indicating the spe-cial character of the Equal Protection Clause and especially the uniquenessof crack sentencing. He then suggests a test for federal sentencing laws,which have disproportionate impact on different populations. AdoptingSklansky’s methodology, I will examine the area of racially based selectiveprosecution. However, I will ªrst consider the coherence argument further.

One of the major tenets of Legal Realism was that generalizations andcategorizations could be misleading. The Realists emphasized the need tolook at the particular situation, and called for narrower categories thatwould be more attentive to particularity and sensitive to context. Oneshould not apply the same rule automatically for the same artiªciallyconstructed category, but instead check whether the rationales and policyarguments that support the rule in one case are also applicable to theother.30 But one does not have to go as far as Realist skepticism to reject theuniversalistic approach adopted by the Court with regard to equal pro-tection jurisprudence. Even Ronald Dworkin, a major proponent of co-

28. David Sklansky, Cocaine, Race, and Equal Protection, 47 Stan. L. Rev. 1283 (1995).

29. 397 U.S. 471 (1970).30. Llewellyn, for example, called for a “narrower, more concrete study.” See Karl N.

Llewellyn, A Realistic Jurisprudence—The Next Step, 30 Colum. L. Rev. 431, 457 (1930)(“Here as throughout we run into the need for reexamining the majestic categories ofthe romantic period of jurisprudence. The old categories are imposing in their pur-ple, but they are all too big to handle. They hold too many heterogeneous items to beof any use.”).


herence, rejects the crude understanding of coherence as an automaticglobal application of rules. Instead, he introduces the notion of “local pri-ority”31 that must be taken into account by the judge together with “ªt” aspart of “coherence.”32

B. Special Features of Criminal Law

(1) The ªrst special characteristic of criminal law is that the implica-tions of discrimination in criminal law are usually much more severe thanin any other area of the law. This is true both from an individualistic pointof view and from a group-oriented perspective. Criminal prosecution mightentail deprivation of liberty, property, or even life. Even in cases that donot result in conviction and punishment, the criminal process has far-reaching inºuences on the individual, such as stigmatization, restrictionof liberty, and emotional suffering. As to the collective point of view, rein-forcing stereotypes and associating a certain race with crime is a way ofperpetuating social inequalities. Therefore, under-detection of discrimi-nation, which is most likely to occur under the current equal protectionstandard, is arguably worse in the criminal justice context than in others.In order to increase the chances of detection, the standard should be morefavorable to the defendant. It should also be directed at protecting indi-vidual rights as well as ªghting group stigmatization.

(2) The criminal justice system suffers from a major problem of legiti-macy.33 Many subjects to the system believe that they are being discrimi-nated against, and large segments of the population distrust the system.This stems partially from the history of criminal law as an oppressive toolused against minority groups,34 a history from which American societyhas not fully recovered. The community’s distrust of the criminal justicesystem, as David Cole points out, has heavy costs in terms of obstructinglaw enforcement, encouraging crime, and deepening societal divisions.35

The intent standard for equal protection challenges does not help thissituation. On the contrary, as Steven Reiss explains, “[W]hen the prose-

31. Ronald Dworkin, Law’s Empire 250–54 (1986).32. The coherence rhetoric can be seen in a different light when looking closely at the

doctrine and discovering how “incoherent” it is in fact. Thus, for example, the testfor peremptory challenges is different and less rigid than the discriminatory intenttest of Davis. See Baston v. Kentucky, 476 U.S. 79 (1986). Sheila Foster argues that “de-spite the equal protection doctrine universalistic ambitions, the Court’s applicationof the discriminatory intent requirement has been far from coherent.” See Sheila Fos-ter, Intent and Incoherence, 72 Tul. L. Rev. 1066, 1069 (1998). See also the view of Duncan

Kennedy, A Critique of Adjudication 33–34, 233 (1997) (“the method of coherencepermits the judge to do ideological work when he furthers a particular legal regimeby developing it in the face of a gap, conºict or ambiguity,” and “constitutional lawtoday is no more coherent than is the common law. However the textual provisionsmay have seemed at the start, the process of interpretation has turned them into ahodgepodge, with some built into particular liberal or conservative agendas, othersdeployed in alteration by liberals and conservatives, depending on which domain ofideological controversy is in question.”

33. For a survey of public opinion polls on that issue see William J. Stuntz, Race, Classand Drugs, 98 Colum. L. Rev. 1795, 1797 n.6 (1998).

34. See generally Randall Kennedy, Race, Crime and the Law 29–136 (1997).35. David Cole, No Equal Justice: Race and Class in the American Criminal Jus-

tice System 169–80 (1999).


cutor’s mental state is the fulcrum of the constitutional restrictions on heractions . . . a defendant will believe that her fate is largely tied to theprosecutor’s subjective feelings towards her.”36 A non intent-based objec-tive test is likely to reduce the level of skepticism of minority groups to-ward the criminal justice system.

(3) An equal protection challenger in criminal cases usually seeks toavoid the charges, whereas in many other non-criminal cases she seeks topositively attain some beneªts.37 Providing beneªts to a group of peoplethat has been discriminated against can be very expensive. This could bean argument—although a highly contested one—for maintaining the highhurdle for equal protection challenges. By contrast, in the criminal justicecontext, assuming that the remedy for a successful equal protection chal-lenge is dismissal of charges, these costs are avoided.

One may argue that, although dismissal of charges does not involveªnancial costs, it involves other costs to society, namely, the cost of havingguilty defendants go free. Karlan is only partially right when explainingthat here, unlike other cases of dismissal, the unpleasant feeling of pro-viding a windfall to a guilty defendant that could have been prosecutedbut for the “mistake,” is avoided. According to her, the assumption is thatthe defendant was singled out, and would not have been prosecuted savefor the equal protection violation.38 But this is only a partial response. Insome cases, absent the equal protection violation, both the defendant andothers who were not prosecuted would have been prosecuted.

The other part of my response to the cost argument is normative.These costs are part of the price society has to pay for equality. It is not anoble idea that society ought to be willing to pay some price for guaran-teeing constitutional rights. Many of the criminal law constitutional andprocedural guarantees are based on the assumption that some values, likehuman liberty and human dignity, have to be secured even at the expenseof freeing some guilty people. The most obvious example is the exclu-sionary rule. Equality is not a lesser value than other values for which weare willing to pay this cost. Furthermore, as I argued before, unequaltreatment by the criminal justice system has its own costs in terms of le-gitimacy. These costs outweigh the cost of not prosecuting some guiltypeople.

C. Special Features of Prosecutorial Discretion

James Vorenberg raised the concern that broad and almost uncheckeddiscretion of prosecutors will result in a situation in which “society’s mostfundamental sanctions will be imposed arbitrarily and capriciously andthat the least favored members of the community—racial and ethnic mi-

36. Steven Alan Reiss, Prosecutorial Intent in Constitutional Criminal Procedure, 135 U. Pa.

L. Rev. 1365, 1431 (1987).37. Karlan mentions this difference in her discussion of the possible remedies for selec-

tive prosecution. The Court in Armstrong left the remedy question undecided. Kar-lan’s discussion convincingly concludes that dismissal of the charges is the onlyplausible remedy. See Karlan, supra note 27, at 2027–30.

38. Id.


norities, social outcasts, [and] the poor—will be treated most harshly.”39

Indeed, the breadth of prosecutorial discretion and the fact that courts arevery reluctant to second-guess this discretion is incompatible with the far-reaching implications of this discretion. One would expect that the morepower an administrative agent has to affect people’s lives, liberty, and prop-erty, the more this power will be conªned by clear guidelines and checkedby judicial review. This is, however, not the case with prosecutorial dis-cretion. Applying the requirement of proof of discriminatory intent to se-lective prosecution exacerbates the dangers of under-reviewed prosecutorialpowers.

A comparison between prosecutorial discretion and decisions of othergovernmental bodies (mainly those of legislatures) that are subject to thediscriminatory intention equal protection standard suggests that there areseveral features of prosecutorial discretion that support adopting an al-tered standard for equal protection challenges in this context.

(1) Prosecutors usually operate away from the public eye. In compari-son to the legislative process, prosecutorial decision-making is also muchless well-documented. Identifying discriminatory intent under these con-ditions is more difªcult.40

(2) While legislation usually uses general terms (e.g., “the federal pun-ishment for crack trafªckers is a minimum of x years imprisonment”),prosecutorial work, in many cases, involves particular decisions with nogeneralization (e.g., “x will be charged with crack trafªcking in a federalcourt”). Thus, it would generally be easier to track discriminatory intentin legislation, based on the generalizations it uses, than in the particulardecisions of a prosecutor.41

(3) In contrast to the legislation process, individuals often make prose-cutorial decisions. Presuming that the majority of the population is notracist, the probability that a single prosecutor will be racist is higher thanthe probability that there will be a racist majority among a group of leg-islators. In addition, the dynamics of group decision might reduce the riskof invidious decisions. This is another reason to suspect that prosecutorialdiscretion is more prone to yield discriminatory decisions.

(4) In the criminal justice system in the United States, the prosecutoris the defendant’s adversary. This might result in over-zealousness and

39. James Vorenberg, Decent Restraint of Prosecutorial Power, 94 Harv. L. Rev. 1521, 1555.

(1981).

40. It is true, however, that attributing intentions to legislative bodies and identifyingthese intentions invoke complex theoretical questions. See, e.g., Jeremy Waldron, Leg-islators’ Intentions and Unintentional Legislation, in Law and Interpretation: Essays

in Legal Philosophy 329 (Andrei Marmor ed., 1995).41. It is true that prosecutorial decisions are not of a uniform character, and while some

are “particular,” others are more “general.” The target of an equal protection chal-lenge can be, for example, a prima facie neutral prosecutorial policy (e.g., to chargeonly inner cities’ residents with crack trafªcking). Indeed, in some cases the work ofthe prosecution resembles that of the legislature, and in those cases it might be truethat the above argument is less convincing, and therefore the same standard of equalprotection should apply. However, most daily prosecutorial decisions, e.g., whetherto charge someone with a petty offence or to drop the charge, are of the “particular”kind, and therefore more susceptible to unidentiªed discriminatory intent than leg-islative ones.


vigor on the part of prosecutors, who, while trying to fulªll their vocation,might not be critical enough toward their own or their colleagues’ deci-sions. Once a decision to charge someone is taken, a prosecutor might betoo focused on how to win the case, or how to “achieve” the maximumpunishment, and less aware of problems like unconscious racism. Otheradministrative agents, who are not rivals of the claimant, may be moreopen to hear her claims of discrimination and to double-check their owndecisions. Again, this argument suggests that discrimination may occurmore often in the course of exercising prosecutorial discretion.

A possible objection to the proposed analysis is that prosecutors aremore trustworthy and respectable than other administrative agents, andthat it is reasonable to believe that discrimination is less likely to takeplace when they exercise their discretion. Furthermore, it could be arguedthat, because of institutional competence and separation of powers con-siderations, it is better to keep in place the intent requirement, rather thanreplace it with a test that would entail more judicial intervention intoprosecutorial decision-making. These objections should be rejected, forthe following reasons:

(i) As demonstrated in Part I.B, empirical studies afªrm the suspicionthat prosecutors do discriminate.42 These results are not surprising.McAdams elaborates on two reasons that might motivate prosecutors todiscriminate. The ªrst is pure racism. There is no logic in assuming thatthis widespread social disease will skip prosecutors, and the principal-agent problem might prevent the detection of a racist prosecutor. The sec-ond reason is that prosecutors can increase their win rate by targeting ra-cial minorities. The prospect of jury discrimination, for example, might bea motivation for conviction-maximizing prosecutors to selectively prose-cute racial minorities.43 Other reasons that could lead to the unequal treat-ment of cases include politics, public opinion, and bureaucracy. As Radeletand Pierce note:

If a murder of a white has different effect on the bureaucratic andpolitical situation than the murder of a black, as it would if mur-ders of white victims are more publicized than murders withblack victims or perceived as more threatening by politically pow-erful groups, racism will enter the legal system through the prose-cutor’s ofªce even if the prosecutor never explicitly attends to race.

Similar considerations can obviously involve the race of the defendants.44

To these structural reasons, which mostly relate to the prosecutors’ inter-ests, we should add the problem of unconscious racism, which is anotherreason to believe that prosecutors exercise discrimination. It is likely that

42. See also Developments in the Law: Race and the Criminal Process, 101 Harv. L. Rev. 1472,

1525–31 (1988); P. S. Kane, Comment, Why Have You Singled Me Out? The Use of Prose-cutorial Discretion for Selective Prosecution, 67 Tul. L. Rev. 2293, 2295–2300 (1993).

43. See McAdams, supra note 27, at 642–52. Back in the 1970s, Myers and Hagan con-cluded that in a process they labeled “strong case typiªcation” prosecutors allocaterecourses “so as to maximize the ration of convictions (and sometime harsh sen-tences) to manpower invested” See Martha A. Myers & John Hagan, Private and Pub-lic Trouble: Prosecutors and the Allocation of Court Resources, 26 Soc. Probs. 439 (1979).

44. Radelet & Pierce, supra note 11, at 617.


unconscious racism inºuences a prosecutor even more than it affects oth-ers: since prosecutors are often involved with criminals who are membersof minority groups, they are more prone to develop prejudices that wouldunconsciously effect their decisions.

(ii) Separation of powers considerations support a relatively independ-ent prosecution. However, prosecutors are not free to violate the Consti-tution. If we are convinced that violations of the Equal Protection Clausewill be more detectable by changing the current standard for equal pro-tection challenges, or that equality demands a different standard, separa-tion of powers considerations cannot prevail. This is the justiªca-tion for astructure of checks and balances. Providing the courts with an effectivetool to ensure that the prosecution is not violating the Constitution is notcontradictory to the American version of separation of powers. Moreover,the democratic majoritarian argument that calls for restrained judicial ac-tivism is stronger when the judiciary reviews legislation than when it re-views administrative decisions. When the question is the application of alaw by prosecutors, the democratic argument would support interventionof the courts in order to make sure that it is equally applied, which is pre-sumably what the democratically elected legislature intended.

As Reiss notes, the court is no less competent to regulate the prosecu-tion than it is competent to regulate the police, as it does occasionally un-der the current legal regime. More importantly, shifting the equal protec-tion standard from an intent-based one to a more objective one does notnecessarily mean more intrusive intervention in the work of the prosecu-tion.45 The question is what does “prosecutorial independence” mean? Itis clear that my proposal will draw some limits on prosecutorial discre-tion, but one might argue that the need to inquire into inner motives andintentions under the discriminatory intent standard is more intrusive thanis a check of objective measures like the ones I am about to suggest.

For all these reasons it is questionable whether the Court was right instating that “the decision to prosecute is particularly ill suited to judicialreview.”46 However, it is unquestionable that the judiciary has a responsi-bility to protect individuals from being selectively prosecuted, and thatthe current protection comes very close to no protection at all.

D. Special Features of Race Issues

(1) Racism in the United States is a phenomenon that one cannot over-look. But although racism exists, it is surely not something of which peo-ple are proud. Presently, it is very hard to ªnd someone who will admitthat she is racist, or who will openly say, “I think black people are crimi-nals.” A racist prosecutor will most probably conceal her views, leavingno evidence of the fact that her decisions have anything to do with race. Itis therefore probable that evidence which shows that a prosecution wasmotivated by an intention to discriminate against people who exercisedtheir First Amendment rights, for example, will be found more easily thanevidence of racially based discriminatory intent. While the intent require-ment will not prevent some people from proving their non-racially based

45. Reiss, supra note 36, at 1441.46. Wayte, supra note 22, at 607.


equal protection challenge, it will probably be an insurmountable obstaclefor the vast majority of race discrimination challenges.

(2) In his comprehensive classic work on unconscious racism, CharlesLawrence explained that the dichotomy between intentional-and-unconsti-tutional actions and unintentional-and-constitutional actions is a falseone. In his words:

Traditional notions of intent do not reºect the fact that decisionsabout racial matters are inºuenced in large part by factors that canbe characterized as neither intentional—in the sense that certainoutcomes are self-consciously sought—nor unintentional—in thesense that the outcomes are random, fortuitous, and uninºuencedby the decision maker’s beliefs, desires, and wishes.

Americans share a common historical and cultural heritage inwhich racism has played and still plays a dominant role. Becauseof this shared experience, we also inevitably share many ideas, at-titudes, and beliefs that attach signiªcance to an individual’s raceand induce negative feelings and opinions about nonwhites. To theextent that this cultural belief system has inºuenced all of us, weare all racists. At the same time, most of us are unaware of our ra-cism. We do not recognize the ways in which our cultural experi-ence has inºuenced our beliefs about race or the occasions on whichthose beliefs affect our actions. In other words, a large part of thebehavior that produces racial discrimination is inºuenced by un-conscious racial motivation.47

Thus, there is a need to transcend the intent requirement in issues of race.(3) Because races are large groups of people, it is possible to conduct

empirical studies and to collect enough data that will show disparate ra-cial impact in a relatively accurate way. There has been some indicationthat the courts might be willing to consider complete statistical analysesto show evidence of discriminatory impact that can be used to infer dis-criminatory intent.48 This supports the idea that there are plausible alter-natives to the intent requirement, at least in this context.

E. Conclusion of Part II

The intent requirement in equal protection jurisprudence has deªcien-cies that many others have discussed before. In this Part, I have tried toshow that, in the speciªc context of race and selective prosecutions, thisrequirement is even more problematic than it is in general.

The special features of race and criminal law combined are more thanthe sum of the features of criminal law separately and race separately. Inthe United States, the very combination of race and criminal law carrieswith it an additional special meaning—a meaning of subjugation. Whencriticizing the intent requirement, Professor Laurence Tribe has called foradopting an “anti-subjugation” principle instead of the “anti-discrimina-

47. Lawrence, supra note 20, at 322.48. See Developments in the Law, supra note 42, at 1543.


tion” principle that governs equal protection jurisprudence.49 Under theanti-subjugation principle, the focus is not on the perpetuator but on thelegal order and the victims. According to Tribe, this principle is the cor-rect interpretation of the Equal Protection Clause in light of its purposeand history. If Tribe is right with regard to only one area of equal protec-tion, this area must be where race and criminal law intersect.

When prosecutorial discretion with all its unique features is added tothis mélange of race and criminal law, the conclusion must be that weshould relinquish the intent requirement in favor of a test that will supplythe desired protection.

III. The Alternatives—“Pure Impact” and “Unequal Treatment”

A. Rejecting “Pure Impact”

The main reason that the Court holds on to the intent requirement, de-spite the critiques, is most probably the lack of practical alternatives. Thealternative that is usually presented as the only possible one is “dispro-portionate impact.”50 According to this standard, the mere showing that alaw or an administrative action, racially neutral on its face, has a dispa-rate impact on a protected group would be enough to subject it to strictscrutiny, and therefore most likely to count as a violation of the Equal Pro-tection Clause. According to this rule intents are irrelevant. This was in-deed the test that was rejected by the Court in Washington v. Davis. If itwere applied the mere fact that blacks had a signiªcantly higher failurerate than whites would have counted as racial discrimination. Similarly,in the criminal law context, a pure impact test would have enabled anyblack defendant to establish an equal protection claim by showing thatthe likelihood of blacks to be accused of a speciªc offence is higher thanthe likelihood of whites to be accused of the same offence.

The Court’s fear of such a “pure impact” standard is understandable.As I previously mentioned, impact is everywhere. This is as true ofAmerican society as it is of every other society. The unfortunate reality ofour historical conditions is that segments of the population—certainlyminority groups, but also other groups, such as women—are disadvan-taged. Financial, educational and other gaps are sufªcient reasons to sus-pect that many rules and administrative decisions run the risk of having adisproportionate impact. As Tribe notes, “In Washington v. Davis the Su-preme Court feared that adoption of a disparate impact test for equal pro-tection analysis would threaten a whole panoply of socioeconomic and ªscalmeasures that inevitably burden the average poor black more than theaverage afºuent white.”51

Criminal law is no exception. If, for example, it were enough to provethat more blacks than whites were being charged under a certain law inorder to declare that law unconstitutional, too many laws would havebeen declared void, and the criminal justice system could no longer func-

49. Laurence H. Tribe, American Constitutional Law 1502–21 (2d ed. 1988).50. See, e.g., id.; Erwin Chemerinski, Constitutional Law: Principles and Policies

644–45 (2 ed. 2002).51. Tribe, supra note 49, at 1519.


tion. The famous sarcastic comment by Justice Brennan in his dissent inMcClescky v. Kemp, that the opinion of the Court suggests “a fear from toomuch justice”52 is probably somewhat unfair. The fear of the Court is notof too much justice, but of a situation in which the criminal justice systemis not functioning. The pure impact test would have been adequate only ifthe presumption of the Court of Appeals in Armstrong that “people fromall races commit all type of crimes”53 were correct, and if they were com-mitting all the crimes at equal rates. The Supreme Court in Armstrongjustiªably rejected this presumption.54 There are many reasons to assumethat there will be disparity in crime rates between different groups. Themost obvious one is that poor people commit more “street crimes” whilerich people commit more “white-collar” crimes. Since there is a correlationbetween poverty and racial minority groups, there will be a disproportionateracial impact.

The disparate impact test is not only impractical, but it is also unableto accurately detect unequal treatment. Indeed, the disparate impactstandard would have been a good litmus test for discrimination only ifthe “racial irrelevance thesis,” which posits that race is unrelated to crimi-nality, were true. However, the racial irrelevance thesis is probably wrong.Recent scholarly work successfully discredits it, and the conclusionsdrawn from empirical data on which it is based.55 If we acknowledge thatthere are different crime rates among different races, we should agree thatdisparate impact—for example the fact that people of one race are prose-cuted at a higher rate than people of another race—is not proof of dis-crimination.56

B. Unequal Treatment and Similarly Situated Individuals

Is pure impact the only alternative to the intent requirement? I believethat it is not. In fact, the Court in Armstrong held that a different kind of“effect” test—the similarly situated individuals [hereinafter, SSI] test—isone of the two necessary prongs of the selective prosecution standard (theother prong being the discriminatory intent requirement). My claim isthat this ªrst prong, modiªed in ways I will describe later, should be thestandard that a defendant must meet in order to prove her selectiveprosecution claim, without the need to show intent at all. A black defen-dant will need to show that at least one other similarly situated white in-dividual was not prosecuted. The prosecution will, of course, have the

52. 481 U.S. 279, 339 (1987) (Brennan, J., dissenting).53. See Armstrong 517 U.S. at 469.54. Id. at 470.55. See most recently R. Richard Banks, Racial Proªling and the (Ir)relevance of Race, 55

Stan. L. Rev. (showing the ºaws in the claim that “equal hit rates” in law enforce-ment stop-and-search practices prove equal crime rates and therefore support the ra-cial irrelevance thesis, and claiming that the empirical studies are more consistentwith the hypothesis of a race-criminality link) (forthcoming June 2003).

56. This, of course, does not mean that such a situation is satisfactory. One could claimthat the criminal prohibitions that have such disparate impact are problematic andshould be abolished. In addition, or alternatively, one could express dissatisfactionwith the social conditions that create such disparity. However, as I explained before,these issues are beyond the scope of this Article, which deals with the problem ofunequal treatment by prosecutors.


opportunity to dispute the fact that those who were not prosecuted weresimilarly situated. If the defendant succeeds in demonstrating that a SSIfrom a different race was not prosecuted, she will prevail (unless theprosecution can demonstrate by statistical evidence, as I will describe inthe next part, that there is no correlation between the decision to prose-cute and the race of the defendant). The SSI standard is different from thepure impact standard because it does not focus on the general likelihoodof blacks and of whites to be prosecuted, but on whether blacks andwhites who are similarly situated are treated similarly. Unlike the pure im-pact test, which disregards the question whether, the different popula-tions are similarly situated (i.e., commit the same crimes), this SSI stan-dard posits this question at the heart of the examination. The SSI standardhowever shares with the disparate impact test its regard of intentions asirrelevant, and in this sense it is radically different from the discrimina-tory intent standard.

The SSI standard overcomes the main deªciencies of both the pureimpact and the discriminatory intention tests by putting emphasis on thevalues of equality and anti-discrimination, understood as anti–unequal–treatment.57

As previously mentioned, the pure impact test is inadequate becausethere can be many reasons, which are considered to be legitimate, that createsuch impacts. This is not true for the situations that the SSI standard cap-tures. We might accept a situation in which 90% of the charges accordingto a certain statute are brought against blacks, because it might be the casethat 90% of the offenders are blacks.58 But we should certainly not acceptthis situation if all black offenders are being charged, while other simi-larly situated white individuals are not. Simply put, this latter situation isan unequal treatment or discrimination in its most obvious form, and, there-fore, it should be regarded as an Equal Protection violation.

Furthermore, as opposed to the pure impact test, the SSI test shouldnot raise the fear that the criminal justice system will collapse. It is un-likely that a situation in which some are prosecuted, but not others whoare similarly situated and are of a different race, is as common as the phe-nomenon of disparate impact. In any event, in those cases which areproven to be discriminatory under the SSI test, it is better that the dis-crimination be exposed and amended than that it be swept under the rugfor the sake of the appearance of justice and legitimacy.

57. I will use the term “discrimination” from this point on to indicate a different under-

standing from the Court’s understanding of discrimination as a concept that requiresintent. In my opinion, as I will further explain, discrimination occurs in situations inwhich there is unequal treatment of equal persons. Therefore, the anti-discriminationprinciple I am calling for, is not the anti-discrimination principle that Prof. Tribe criti-cized, but it is more similar to the “anti-subjugation” principle that he proposed. Seesupra text accompanying note 49.

58. I intentionally use the words “might accept” as I bracket, for the purpose of this Arti-cle, the problem of criminalization of “black behavior.” For examples of good discus-sions of this problem, see Sklansky, supra note 28; Dorothy E. Roberts, Punishing DrugAddicts Who Have Babies: Women of Color, Equality and the Right of Privacy, 104 Harv L.

Rev. 1419 (1991); Dorothy E. Roberts, Unshackling Black Motherhood, 95 Mich. L. Rev.

938 (1997).


The SSI standard also does not suffer from the main three disadvan-tages of the intent requirement:

(1) It does not fail to capture inequalities and biases that exist regardlessof discriminatory intentions. The SSI test directly focuses on material re-ality rather than on consciousness, and is aimed at ameliorating inequali-ties and structural biases that subsist in the criminal justice system, inde-pendently of any inquiry into a state of mind.

My proposal is different from other suggestions, according to which,after a prima facie showing by the defendant, the burden shifts to the Gov-ernment to show that there was no discriminatory intention.59 I propose toentirely depart from intentions. The burden under my proposal may shiftto the prosecution only to prove that the other individuals are not simi-larly situated (e.g., they committed a less severe crime). In the secondstage of the SSI standard, the prosecution will also be able to introducestatistical evidence that will show that, actually, there is no unequaltreatment. Intentions will remain irrelevant. This might have many impli-cations. For instance, if the prosecution can show that the suspects fromthe other race were negligently not prosecuted, although they were simi-larly situated, according to my suggestion, there is a prima facie violationof the Equal Protection Clause. The emphasis here is on the inequalitythat was caused. From the defendant’s point of view, and from other peo-ple of her race’s point of view, it is of little difference whether the prose-cutor was racist or negligent.

(2) The SSI standard does not ignore the phenomenon of unconsciousracism. Because of the focus on real world inequalities, cases of uncon-scious racism will be addressed and amended. If, for example, a chargewere brought against a black defendant for a petty offence, and the samecharge is dropped when similarly situated white suspects are involved ina similar conduct, it would be enough to pass the ªrst stage of the equalprotection claim, regardless of the fact that the prosecutor’s action wasnot consciously driven by a racist motivation.

(3) As I will show in Part VI, after describing the standard in detail,meeting the SSI standard is not an impossible mission. Clearly, it is aneasier task than proving discriminatory intent.

At this point, one might raise various material questions regarding theSSI standard, which remain to be answered. The rest of this Article, which

59. Thus, for example, the complicated rules controlling Title VII litigation still maintain

the intent element. Plaintiff must identify a particular employment practice that has adisparate impact and the employer bears the burden of persuading the fact ªnderthat the employment practice that has a disparate impact is justiªed by a legitimate,independent, and nondiscriminatory business reason. In other words, that there is nodiscriminatory intention. See 42 U.S.C. § 2000e-2(k) (2003); Tex. Dep’t of Cmty. Affairsv. Burdine, 450 U.S. 248, 252–53 (1981); McDonnell Douglas Corp. v. Green, 411 U.S.792, 802, 804–05 (1973). Similarly, under the standard for examining discrimination inperemptory challenges in jury selection, although different from the conventionalEqual Protection discriminatory intent standard, the intent requirement is not com-pletely abandoned. Defendant must establish a prima facie case of race discrimina-tion. After defendant shows that circumstances raise an inference of discrimination,the burden shifts to the prosecution to offer a race-neutral justiªcation for its per-emptory challenge. The trial judge then determines whether the state’s proffered rea-son is race-based and improper. See Batson v. Kentucky, 476 U.S. 79, 94–98 (1986).


raises and addresses these possible questions, unfolds in the followingway: In the next part, I will reªne the proposed SSI standard. I will ad-dress the question of what number of SSIs a defendant must show in or-der to move to the second stage of her selective prosecution claim. I willalso explain the need for and the nature of the second stage in which theprosecution would be able to present statistical evidence. In Part V, I willanswer the question: who is “similarly situated?” In other words, whatcharacteristics and what level of similarity will sufªce to conclude that twoindividuals are “similarly situated?” What differences between individu-als will be considered as relevant and, therefore, will be enough to refutethe claim that an individual is similarly situated? In Part VI, before con-cluding, I will argue that under the SSI standard, selective prosecutionclaims will be possible to prove, and I explain the reasons for this proposi-tion.

One would expect to ªnd an abundance of legal literature and courtdecisions regarding these questions. The SSI is not a new concept. Quitethe contrary; as previously indicated, it is one of the two prongs of theexisting legal standard for a claim of selective prosecution, and it is alsoone of the conditions a defendant should meet in order to obtain discov-ery on this issue. Yet, the existence of the second prong—the discrimina-tory intent—and the courts’ insistence on its strict interpretation and ap-plication have precluded a robust discussion of these issues. Almost everyselective prosecution claim that was brought to the courts has been dis-missed on the grounds of lack of proof of discriminatory intent. Givingup this last requirement, as I suggest, will place the prescribed questionscenter stage. Although the questions presented above are intertwined, Iwill try to discuss them separately, in the order presented, while indicat-ing their mutual inºuences when appropriate.

IV. Reªning the SSI Standard: Groups, Individuals, and

Statistical Evidence

The standard I am advocating is a twofold test. At the ªrst stage, a de-fendant will have to show that a single SSI was not prosecuted. Such ashowing by the defendant (that was not refuted by the prosecution) willshift the burden to the prosecution to disclose the full data with regard topeople who were involved in conduct similar to the defendant’s conduct,including racial proªles of suspects who were and were not prosecuted.The prosecution will have to show, using these data, that there has beenno discriminatory practice. I will later explain what will be considered asa full disclosure of the relevant data. A failure to disclose all relevant data,or a failure of the data to demonstrate that there is no correlation betweenthe race of the defendant and the decision to prosecute, will result in vic-tory for the defendant.

Each one of the two proposed stages might at ªrst glance appear to besurprising. I will explain the policy arguments that support this twofoldstandard, show that the two stages together create a logical structure, andargue that the proposed test is compatible with the theoretical critiquesthat have been launched against the discriminatory intent standard.


A. The First Stage—A Single SSI

A situation where a black defendant was prosecuted, while a whiteperson similarly situated to her was not prosecuted, calls for our suspi-cion. Allegedly, it is exactly what the proposed standard is aimed atmending, i.e., unequal treatment by the criminal justice system. We are nolonger interested in the state of mind of the prosecutor. It could be thatthere was intentional discrimination, unconscious discrimination, or nei-ther of these two. What needs to be remedied is the de facto situation inwhich individuals from different races are treated differently. One whothinks historically and contextually about criminal justice in the UnitedStates cannot ignore the singularity of racial discrimination. Therefore,despite the fact that a situation in which only one SSI was not prosecutedmight be explained as a “mistake” or as a reasonable decision based onlegitimate grounds, it is enough to trigger ample legal litigation, whichwill beneªt not only the particular defendant, but also the whole system.

There is one court decision that might look at ªrst glance to be in op-position to the idea that a single SSI is enough. This decision is in factsometimes cited as a source for denial of selective prosecution claims onthe grounds that a showing of only a small number of offenders who werenot prosecuted is meaningless.60 This understanding of the case is mistaken.

In United States Labor Party v. Oremus,61 the Labor Party claimed that theIllinois statute prohibiting solicitation in intersections was discriminato-rily enforced. In support of its claim, an afªant asserted that he observedmembers of the Lions Club soliciting in the same intersection as the LaborParty did. The Court of Appeals for the Seventh Circuit rejected the equalprotection claim, indicating that “the isolated incident of another groupbrieºy soliciting in the intersection is too insigniªcant and isolated to raisean equal protection claim.”62 This sentence is probably the cause for themisunderstanding of United States Labor Party as attributing no signiª-cance to a showing of a single case of an SSI who was not prosecuted.

A close reading of the case reveals that in fact this was not the basisfor the rejection of the equal protection claim. The court did not hold thata showing of only one incident of SSI is not enough. Rather, the single in-cident that was presented was not of the same kind as the Labor Partyincident. As the court explained:

The plaintiffs [the Labor Party] at least had the advantage of inter-section solicitation, although illegal, for several weeks whereas theLions Club group was observed in the intersection only for a sin-gle hour period. Perhaps, if anyone might have cause to complainin those circumstances it would be the Lions Club . . . .63

Thus, the Court held that the Lions Club was not similarly situated tothe Labor Party, and distinguished between them on the basis of the ex-tent of their illegal activity.64 Therefore, The Labor Party cannot stand for

60. See, e.g., Reiss, supra note 36 at 1371–72 n.21.61. 619 F. 2d 683 (7th Cir. 1980).62. Id. at 691.63. Id.64. See discussion regarding classiªcations related to the criminal conduct in part V(D).


the proposition that one has always to show more than one SSI. In any case,my argument is that indeed a single SSI is not enough to win the equalprotection claim, but that it is enough to trigger a more thorough investi-gation of the matter.65

B. The Second Stage—Prosecution’s Burden To Provide CompleteStatistical Data

A showing of a similarly situated individual who was not prosecutedreveals that, prima facie, something went wrong. Still, further inquiry isneeded because a showing of a single SSI does not necessarily represent asituation in which there is a correlation between the race of the defendantand the decision to prosecute. In a system such as the criminal justice sys-tem, a single case could be a mistake, one that does not represent such acorrelation. In fact, if we are interested in changing the de facto situationsin which one racial group is treated differently than another (be it becauseof intentional discrimination, unconscious discrimination, structural bias,or any other reason), we must formulate a statistical picture that will in-clude four variables: the number of SSI white people who were prose-cuted, the number of SSI white people who were not prosecuted, the num-ber of SSI black people who were prosecuted, and the number of SSIblack people who were not prosecuted.66 As I will demonstrate below, ifone or more of the four variables are missing, there is a danger of arrivingat an incorrect conclusion.

The danger in presenting a partial statistical picture has, at times, beenignored by the courts. There are two paradigmatic cases that can demon-strate this danger. In each of these cases, one might erroneously concludethat a partial showing by the prosecution is enough to dismiss the blackdefendant’s claim of selective prosecution, and to regard as meaninglessher showing that a similarly situated white individual was not prose-cuted. The two paradigmatic cases are: (1) When there are other similarlysituated people, not of the defendant’s race, who were prosecuted; and(2) When there are other similarly situated people, of the defendant’srace, who were not prosecuted.

(1) An example for the ªrst paradigmatic case can be found in State v.Holloway.67 The defendant was indicted on two counts of evasion of per-sonal state income tax and two counts of ªling false statements. He ªled amotion to dismiss on the grounds of selective prosecution, arguing that

65. Other cases that might support the position that a showing of only one SSI is mean-

ingless are: Dawson v. City of New York, No. CIV.97-5347-TPG, 1999 WL 46624, at *1(S.D.N.Y. Feb. 02, 1999) (holding that to prevail on a selective prosecution claim, aplaintiff must establish that others similarly situated were “generally” treated differ-ently due to an improper consideration such as race), vacated by 199 F.3d 1321 (2ndCir. 1999) (citing United States v. White, 972 F.2d 16, 19–20 (2d Cir. 1992)). Requiringthe defendant to prove that “generally” SSIs were treated differently is unreasonable,because it is hard to imagine how someone can ever prove such a generality. Re-quiring the prosecution to present comprehensive data, as does the second stage ofmy proposal, is a way to address the generality concern.

66. See McAdams, supra note 27, at 627–28. The structure of the Tables that I will use isborrowed from McAdams.

67. 460 A.2d 976 (Del. Super. Ct. 1983).


he was a target for prosecution because of his racial and political status asa black Representative in the General Assembly of the State of Delaware.The defendant further argued that he is the ªrst and only one to be prose-cuted under these kinds of accounts, separate from other alleged criminalconduct. The state in response cited ªve other cases in which individualswere prosecuted on the same charges.68 Rejecting the defendant’s claim,the Court stated: “Inasmuch as a showing of as few as two or three otherprosecutions will negate the assertion that defendant has been singled outfor prosecution, the court holds that the ªrst prong of the Berrios test [re-quiring proof of SSI] has not been satisªed.”69 [Footnote omitted].

What underlies the decision in Holloway is the fact that the added data(two or three individuals that were prosecuted) changes the probabilitiesof prosecutions, and, therefore, it might affect the conclusion regardingthe question of whether there is a correlation between the race of the de-fendant and the decision to prosecute. Take as an example a case in whicha black defendant claims she was singled out for prosecution on the basisof her race, and supports her claim in a showing of a white SSI who wasnot prosecuted. If this is all the data we have, one may assume that theprobability of a black person to be prosecuted is 1 while the probability ofa white person to be prosecuted is 0. Assuming the prosecution can showthat there were 10 other whites similarly situated, who were prosecuted,the total picture would change and will look as follows:

Blacks Whites

Prosecuted (A) 1 (C) 10

Not Prosecuted (B) 0 (D) 1

The probability that a black person will be prosecuted is the numberof blacks who were prosecuted (cell A) divided by the sum of cell A plusthe number of blacks not prosecuted (cell B). The probability of a whiteperson to be prosecuted is the number of whites prosecuted (cell C) di-vided by the sum of cell C plus the number of whites not prosecuted (cellD). With the above hypothetical data, the probability of a black person tobe prosecuted is: A / (A+B) = 1 / (1+0) = 1, and the probability of awhite person to be prosecuted is: C / (C+D) = 10 / (10+1) = ~0.909. Theadditional data have changed the picture dramatically in a way that, tosay the least, weakens the black defendant’s claim.

It is clear, then, that a showing of a single SSI by the defendant shouldnot be the ªnal step in proving unequal treatment. As the example indi-cates, a showing of other individuals, outside of the defendant’s protectedgroup, and who were prosecuted, can change the picture. But still, theholding of the court in Holloway that “a showing of as few as two or three

68. An important question here is, are these ªve “similarly situated?” The defendant

claimed that they are not similarly situated because their indictments contained ad-ditional charges. The court rejected this argument. For a discussion and a critique ofthe way the courts determine similarities and dissimilarities see infra Part V. See es-pecially Part V.D, in which I discuss this aspect of Holloway.

69. Holloway 460 A.2d at 979.


other prosecutions will negate the assertion that defendant has been sin-gled out for prosecution” is clearly wrong if that means denial of theequal protection claim. Why two or three? Why not one or four? Indeed,any such number is arbitrary and meaningless, unless it is accompaniedby the rest of the data regarding the numbers of blacks and whites whowere prosecuted, and who were not prosecuted.

Another example might be useful here. Let us assume that after ashowing by a black defendant, that a similarly situated white person wasnot prosecuted, the prosecution shows that ªve other similarly situatedwhite individuals were prosecuted. According to the opinion of the courtin Holloway, this showing will negate the defendant’s claim of selectiveprosecution. The problem with this conclusion is that it relies on incom-plete data (exactly as any conclusion drawn from the defendant’s show-ing of a single SSI). Taking this new data into account, the probability of awhite individual to be prosecuted is 5/6, and one may argue that it is notsigniªcantly different than the probability of blacks to be prosecuted(1/1). But what if there were, for example, 50 other white people whowere not prosecuted? In this case the statistical picture will be:

Blacks Whites



The probability of a black person to be prosecuted is: 1 / (1+0) = 1,and the probability of a white person to be prosecuted is: 5 / (5+51) =~0.089. This means that a black person has more than ten times the chanceto be prosecuted than a white person has. This is undeniably and consid-erably different from the difference between 1 and 5/6, and might changeour ªnal conclusion regarding the merits of the selective prosecutionclaim.

(2) Similar difªculties arise in the second paradigmatic case, i.e., whenother members of the protected group of the defendant were not prose-cuted. Here, too, some courts fell into a logical trap similar to the one intowhich the court in Holloway fell.70 If we go back to our example (in whichone black was prosecuted and one white was not), and add to it a show-ing by the prosecution of 10 blacks who were not prosecuted, it willchange the picture to be as follows:

70. In United States v. Olvis, 97 F.3d 739 (4th Cir. 1996), the Court of Appeals for the 4th

Circuit reversed a dismissal of indictment by the district court on the grounds of re-fusal of the prosecution to comply with the court’s discovery order. The court of ap-peals found that there was no merit to the defendants’ contention that they were sin-gled out for prosecution, while other similarly situated whites were not prosecuted.One of the reasons mentioned by the court of appeals for the lack of merit to defen-dants’ claim was that there were also ªfty black conspirators who were not prose-cuted. Id. at 745


Blacks Whites



The difference between the probabilities of a black person to be prose-cuted (1/11) and a white person to be prosecuted (0/1) is not as dramaticas it was before the added data.

But this showing of 10 blacks that were not prosecuted should notnecessarily be the end of the story. Again, the picture as a whole can betotally different. Such will be the case if, for example, there are 100 moreblacks who were prosecuted:

Blacks Whites



Under the new set of data, the probability of a black person to beprosecuted has increased to 101/111 = ~0.91, while the probability of awhite to be prosecuted has remained 0. The difference between the prob-abilities is again substantial.

Some cases might involve a combination of the two paradigmatic cases.It might be that the prosecution would be able to demonstrate both thatthere were some blacks who were not prosecuted and that there were somewhites who were prosecuted. Again, if these showings are intended toprove that there is no correlation between the decision to prosecute andthe race of the defendant, it is important to understand that such a con-clusion might be ºipped by adding more data, such as that more blackswere prosecuted.

All these examples show that the similarly situated individual shouldonly be a ªrst stage—a trigger that calls for presenting the statistical pic-ture as a whole. It is important not to be misled by the limited availableªgures. As opposed to what the courts held or hinted in some cases,71 “dis-parate treatment” is not necessarily negated by a showing of members ofthe protected group who were not prosecuted, or by a showing of indi-viduals who are not members of the protected group who were prose-cuted. If we open the gate for usage of partial data, we also open it formanipulation. The prosecution, for example, could use some rare cases, inwhich whites were prosecuted, as a ªg leaf that covers and legitimizes

71. See the description of Holloway, supra text accompanying note 67; Olvis 97 F.3d at 745;

See also People v. Goodman, 290 N.E.2d 139, 143 (1972) (“Even though the defendantasserts that certain other offenders have not been prosecuted, the record disclosesthat on the day he was sentenced, three other persons were convicted under thesame ordinance”); National Railroad Passenger v. Harris, 490 F.2d 572, 574 n.1 (10thCir., 1974) (In its reasoning for rejection of the defendants’ claim of selective prosecu-tion, the court mentions the fact that there were approximately thirty-four othersimilar prosecutions in the same county during the nine-month period precedingthat case).


unequal treatment. Immunizing the prosecution against equal protectionchallenges because they can show that some whites were prosecuted, orthat some blacks were not, is a clear mistake. As we have seen above, suchshowings by the prosecution can present a distorted picture of reality.

We cannot know if there is a correlation between the race of the de-fendant and the decision whether to prosecute, unless we have the fullpicture—that is, the four variables represented in the four cells of the ta-bles (A, B, C, D). Only the whole picture can reveal the real situation ofthe enforcement of different statutes in relation to different racial groups.

Let us now demonstrate the application of the formula to real data. Iwill use the data regarding the state and federal prosecution of blacks andwhites for crack cocaine offences in Los Angeles. The question at stake is:are blacks treated unequally in the sense that there is a correlation be-tween their race and the decision to prosecute them in federal courts? Ac-cordingly, I will substitute in the table “prosecuted in federal court” for“prosecuted” and “prosecuted in state court” for “not prosecuted.” Usingthe data that I mentioned before,72 the picture will look as follows:

Blacks Whites

Federal (A) 36 (C) O

State (B) 4410 (D) 222

Assuming that these numbers represent similarly situated individuals,the probabilities to be prosecuted in federal court are:

For blacks—36 / (36+ 4410) = ~ 0.008For whites—0 / (0+222) = 073

Whatever the reason, the very situation of unequal treatment should beregarded as a violation of the Equal Protection Clause. However, the com-plete statistical data that are necessary to prove unequal treatment areusually unattainable for the defendant. The prosecution, on the other hand,has the ability to acquire such information. For that reason, and for thereason that a showing of one SSI already renders the situation suspicious,the burden of providing this data should rest on the prosecution.

To sum up, my suggestion is as follows: after a black defendant hasdemonstrated that there is at least one similarly situated white individualwho was not prosecuted, the prosecution can contradict her claim of se-lective prosecution only by providing complete data of the numbers ofblacks similarly situated who were prosecuted, blacks similarly situatedwho were not prosecuted, whites similarly situated who were prosecuted,

72. See supra text accompanying note 14.73. It is beyond the scope of this Article to address the question of what kind of differ-

ence between the probabilities should be regarded as a strong enough correlationbetween race and the decision to prosecute. Such a determination hs to be based on amore sophisticated statistical model that will take into consideration other factorssuch as standard deviations and the amount of the data. My aim here is only to pres-ent the basic model and to demonstrate its ability to detect situations of unequaltreatment.


and whites similarly situated who were not prosecuted. Having all thesedata, an accurate determination of whether there is a correlation betweenthe decision to prosecute and the race of the defendant can be made. If thedata provides a positive answer to that question, and thus afªrms thesuspicion that was raised by the showing of a single SSI who was not prose-cuted, the defendant should prevail in her selective prosecution claim. Ifthe prosecution fails to provide the data, the suspicion remains intact andthe defendant should also prevail.

C. Scope of the Data: Limits of Jurisdiction and Time

A question that has to be considered is what are the limits of theprosecution’s duty to provide data in terms of time and location? In bothof these aspects, it is impossible and undesirable to compel the prosecu-tion to provide data that is unlimited. Like many other criteria of meas-urement, here too, the exact point at which we draw the line is somewhatarbitrary. Nevertheless, it should be determined.74 I will not propose herea clear rule, because such a rule depends on circumstances that maychange from one place to another. However, I will point out policy con-siderations that should be taken into account in designing such rules.

1. Jurisdiction

When dealing with the evidence to be presented in order to prevail ina discovery motion in a selective prosecution claim, the Supreme Courthas left open the question of whether a nationwide showing (as opposedto a showing regarding the record of the particular decision makers) issufªcient.75

My opinion is that each prosecution ofªce, which usually overlapswith a county, should provide data regarding its own jurisdiction. Thiswill create incentives for each prosecution ofªce to act in a non-discrimina-tory way, and to collect the data efªciently. If we take a broader pool ofdata (e.g., nationwide or statewide data), there will be a problem of exter-nalization of the “costs” of an inadequate behavior. In the Law and Eco-nomics literature, this problem is referred to as “the common pool prob-lem,” which is a speciªc case of “the collective action problem.”76 The in-

74. In Jewish law, there is a fascinating discussion regarding such “standards and meas-

ures”—their arbitrariness on the one hand but the necessity to have them on theother hand: “A ºedgling bird that is found within ªfty cubits [of a dovecote] belongsto the owner of the dovecote; [if it is found] beyond ªfty cubits, it belongs to itsªnder . . . . Rabbi Jeremiah asked ‘What is the law if one foot [of the bird] is withinªfty cubits and the other foot is beyond ªfty cubits?’ On account of this, they ex-cluded Rabi Jeremiah from the study-hall.” Babylonian Talmud, Tractate Bava Batra23b. Rashi explained: “for he caused them bothersome and annoyance.” And theTalmud explains: “This is the character of all measurements established by the Sages.A ritual bath must contain forty se’ah of water to be ªt for immersion; if it containsforty se’ah less one kurtov [a very small liquid measure], it cannot be used for immer-sion.” Babylonian Talmud, Tractate Ketubbot 104a.

75. Bass, 536 U.S. 862.76. A classic article discussing this problem is Garrett Hardin, The Tragedy of Commons,

162 Science 1243 (1968). See also Elinor Ostrom, Self-Governance of Common Pool Re-sources, in The New Palgrave Dictionary of Economics and the Law, vol. 3, 424(Peter Newman ed., 1998).


centive to act adequately in a situation of “common pool” is reduced be-cause each actor’s (or institution’s) behavior does not exclusively deter-mine its own future, but only a part of it. The prosecution ofªce in BronxCounty, New York will have an incentive to not engage in unequal treat-ment so that if a defendant raises an equal protection claim and shows anSSI who was not prosecuted, the ofªce will be able to refute the claim byproviding data kept by the ofªce. However, if the prosecution ofªce in theBronx could be easily affected by the misbehavior of the prosecutionofªce of, say, Harris County (Houston), Texas, or even by another prose-cution ofªce in an upstate New York county, the incentive for the Bronxofªce to keep its own record “clean” will be reduced. Furthermore, a smallprosecution ofªce might not have the incentive to act adequately, becauseunder the common pool of data its racist or negligent behavior might beinsigniªcant. The best incentive will be supplied by holding each ofªceaccountable for its own behavior.77 An ofªce within a county makes senseas a common pool, as it could be reasonably expected that an ofªce,through its senior staff, could ensure equal treatment by generating poli-cies, by supervising, and by taking adequate hiring decisions.78

2. Time

On the one hand, the time limit should provide enough duration toenable a relatively wide perspective. Data regarding the last couple ofmonths, for example, can be biased and not representative of real prac-tices. On the other hand, the time to which the data refer should be lim-ited in such a way that the prosecution will have a chance and an incen-tive to “improve” and change discriminatory practices. This will be achievedonly if the time limit chosen will be short enough to allow new improve-ments to signiªcantly affect the complete picture that ought to be pre-sented.

In Armstrong the discovery order granted by the district court re-quired information regarding three years,79 in Jones the motion was forªve years,80 and in another decision, which granted a discovery order in aselective prosecution claim, the demand was for information regardingthree and a half years.81 Such periods of time are reasonable and are com-patible with the above considerations.

77. In some cases, the issue is such that it must involve several prosecution ofªces. For

example, this is the case when there is a behavior that is criminalized both by stateand federal laws, and the discrimination claim is that the defendant was prosecutedin federal court while other SSI were prosecuted in state courts.

78. The extent to which different prosecutors in different counties can have differentattitudes and policies is exempliªed in the disparity between counties with regard tothe decision to seek the death penalty. See Richard Willing & Gary Fields, Geographyof the Death Penalty, USA Today, Dec. 20, 1999, at 1A; Richard Willing, Prosecutor OftenDetermines Which Way Case Will Go, USA Today, Dec. 20, 1999, at 6A. The “geographi-cal discrimination” evidenced in these newspaper articles is an issue well worthy ofanother article.

79. Armstrong, 517 U.S. 456 at 459.80. Jones, 159 F.3d at 975. The case was remanded to the district court to compel discov-

ery.81. See United States v. Tuitt, 68 F.Supp.2d 4 (D. Mass. 1999).


D. Possible Objections

An immediate objection that comes to mind regarding the proposedtwofold standard could be that it will be relatively easy for a defendant topass the ªrst stage by ªnding only one SSI, and thus to impose on the prose-cution an unbearable burden in almost every case. My response to thisobjection is that it is true that in comparison to the standard of discrimi-natory intent the SSI standard will be easy to prove, but it is not true thatin every case a defendant will be able to meet this standard. As I will ex-plain in the next Part, the prosecution, before having to provide the data,could claim that the defendant is not similarly situated to the other indi-vidual who was not prosecuted, speciªcally, that their criminal behaviorsshould be distinguished. I deªnitely do not view the criminal justice systemthrough rose-colored glasses, especially when it comes to race issues, butI am not so pessimistic as to think that in so many cases, under the juris-diction of so many different prosecution ofªces, in so many differentcontexts, the same pattern of SSIs who were treated differently will recur.

However, in cases where there is an SSI from a different race who wastreated differently, it is justiªable to demand an explanation from the prose-cution for what looks like unequal treatment. The fact that the prosecutionwill have to engage in collecting and presenting statistical data should bewelcomed, rather than feared. There are many beneªts in encouraging theprosecution to conduct studies regarding racial impact, apart from theobvious need to amend the speciªc cases in which unequal treatmentwould be revealed. Awareness of the possibility that in some cases statis-tical information will be required will have a positive effect on the generaldaily attitudes of prosecutors regarding the problem of racial disparity,and thus might reduce the chances of unconscious racism. Moreover, thesubmission of such empirical studies to the courts will make this dataavailable through the media to the public. The prosecutors will thus bepublicly accountable, and the public will be able to express its satisfactionor discontent. In jurisdictions in which prosecutors are elected, the publicwill be able to express its views through the electoral process.82

It might be argued that the prosecution simply does not have the rele-vant data in many cases. After the next part, it will be clear that the datathat are required to be presented under the proposed standard are notcomplicated to collect. Personal circumstances of the different offenders,for example, will not be part of this data. But one may argue that even thenecessary basic data, namely the racial proªle and the criminal activity ofthe offenders, are not attainable because the prosecution simply does notkeep such records. These kinds of claims were raised by the governmentin United States v. Olvis. The court rejected these arguments:

Assistant United States Attorney Robert J. Seidel, Chief of the Nor-folk and Newport News Criminal Division, submitted an afªdavitto the Court in which he indicated to the Court that the UnitedStates Attorney cannot provide any data on the race of defendantsprosecuted. The Court certainly cannot ignore that racial data maybe retrieved from defendant information sheets, which AUSA

82. See Davis, supra note 7, at 54–67.


Comstock acknowledged existed, rap sheets, criminal records, caseagents or sources other than those called to the Court’s attentionby AUSA Seidel. The Court only can assume that these records aremaintained by the United States Attorney, although the numbersmay not appear in an easily produced chart or database.83

Even if the court in Olvis was wrong and the government was rightregarding the current situation, this situation must be changed. The problemof unequal treatment in the criminal justice system is grave enough to re-quire prosecution ofªces all over the country, to keep relevant records andto be able to access them, if it is not presently the case.

The aspect that seems more problematic in the requirement to providedata is that the data should include not only the individuals who wereprosecuted, but also those who were not prosecuted. This might be moredifªcult to comply with, because it may be that some of those who werenot prosecuted were not even arrested, and there is no ªle on them. Again, ifthis is the situation, it calls for a change. It is reasonable to require theprosecution to have a record that indicates each suspect’s behavior andher race. If the reason for not having prosecuted a suspect is insufªciencyof evidence, this individual will not be taken into account. It is worthstressing, however, that this proposal is not revolutionary in terms of thedata the prosecution should keep. In cases in which the court granted adiscovery order under the current legal regime, the data that were re-quired were not less comprehensive than the data that will be required tobe presented under the new standard.84

Another possible objection to my proposal is that the ªrst stage ofshowing SSI will be a hurdle that only a few defendants will be able to pass.Thus, practically, the adoption of the new standard will not create a realchange in the system, and will not be helpful for improving the situationof racial minorities. At the outset, it is somewhat reassuring that the pro-posed standard could be reasonably expected to be challenged by twoopposing camps.85 The fact that I can expect contradictory objections tomy proposal—one claiming that too many people will be able to showSSI, and the other claiming that too few will be able to do so—is a signalthat the proposed standard is actually a balanced one. It will not create acollapse of the criminal justice system or the prosecution, but it will re-quire some changes of practice aimed at a more egalitarian system. How-ever, since I am deeply concerned by the latter objection, I will address itat length in Part VI. Before that, I will discuss in Part V the deªnition of“SSI,” which will have implications for the question of the ability or inabilityto meet the standard.

V. Similarities and Dissimilarities—Who Is Similarly Situated?

The idea that equality concerns the “similarly situated individual” re-lies on the Aristotelian understanding of equality as equal treatment for

83. United States v. Olvis 913 F. Supp. 451, 456 (E.D. Va. 1995), rev’d on other grounds, Ol-

vis, 97 F.3d 739.84. See Armstrong, 517 U.S. 456; Tuitt, 68 F. Supp. 2d 4.85. In talking with friends and colleagues about this proposal, I indeed received those

two critiques.


equals.86 If persons are not equals (or not “similarly situated”), it may bejust to treat them unequally. The accredited notion of “substantive equal-ity” as opposed to “formal equality” is derived from this understanding,and it is the basis of the Equal Protection Clause of the United States Con-stitution, as interpreted by the Supreme Court and by scholars.87

The question I am confronting in this Part is who will be consideredas similarly situated in the speciªc context of a claim of racially based se-lective prosecution? No factual situation is equal in all senses to another,and no individual is equal in all senses to another. Indeed, the ways todistinguish between a black defendant and a white person who is alleg-edly similarly situated and was treated differently could be inªnite. Hence,the requirement is not to show someone “identically situated” but only“similarly situated.” An “identically situated” standard is not only im-possible as a matter of proof, but it is in fact an ontological impossibility.

The challenge here is to deªne “similarity” in a way that will on theone hand be sensitive to the need to treat different people differently, i.e.,to provide a deªnition that will allow the prosecution to distinguish be-tween individuals on the basis of legitimate considerations, but that onthe other hand will not be too narrow so that any form of unequal treat-ment could be justiªed as treating dissimilar people differently. Two ques-tions seek answers: what will be considered as a difference? And whatsimilarity will the defendant have to prove in order to shift the burden tothe prosecution to prove such a difference? I will start with the formerquestion, which is primarily a substantive one, and will then proceed tothe latter, which is more procedural-evidentiary.88

Examination of court decisions in which the question of similarityarose in the context of selective persecution conªrms that there could in-deed be dozens of claims of differences between individuals. Hence, it seemsnecessary to categorize these claims in a way that will encompass all fore-seeable attempts of distinction. The four categories that I ªnd useful for

86. Aristotle, Nicomachean Ethics 71 (Terence Irwin trans., Hackett Publishing 2d

ed. 1999) (1985) (“Equality for the people involved will be the same as for the thingsinvolved, since [in a just arrangement] the relation between the people will be thesame as the relation between the things involved. For if the people involved are notequal, they will not [justly] receive equal shares; indeed, whenever equals receiveunequal shares, or unequals equal shares, in a distribution, that is the source of quar-rels and accusations.”)

87. See, e.g., City of Cleburne v. Cleburne Living Ctr., Inc., 473 U.S. 432, 439 (1985) (“TheEqual Protection Clause of the Fourteenth Amendment commands that no State shall‘deny to any person within its jurisdiction the equal protection of the laws,’ which isessentially a direction that all persons similarly situated should be treated alike.”); 3Ronald D. Rotunda & John E. Nowak, Treatise on Constitutional Law—Sub-

stance and Procedure (3d ed. 1999) 209 (“Equal protection is the guarantee thatsimilar people will be dealt with a similar manner and that people of different cir-cumstances will not be treated as if they were the same”).

88. As one can see from the second question, the stage of showing a single similarlysituated individual will actually be composed of two sub-stages: the showing by thedefendant and the possibility of it being refuted by the prosecution. To be clear, thesetwo sub-stages together are only the ªrst of a twofold process in the selective prose-cution claim. The second stage, the statistical evidence one, will be triggered only ifthe defendant succeeded in his showing of a similarly situated individual, and theprosecution did not succeed in refuting this showing. This part deals with the ªrststage (including both sub-stages).


this purpose are: (a) distinctions that refer to the individual’s being a partof a protected group (or a group that should be protected); (b) class dis-tinctions; (c) other personal circumstances; and (d) circumstances and dis-tinctions that are related to the criminal conduct. I will argue that, whiledifferences of the fourth category might justiªably deny a claim that twoindividuals are “similarly situated,” differences of the ªrst, the second,and the third groups should not be regarded as “differences that make adifference” in the limited context of the ªrst stage of the proposed stan-dard. Nevertheless, there are important distinctions between the ªrst, thesecond, and the third categories that I will explore below.

A. Suspect Classiªcations (and Classiªcations That Should Be Suspected)

1. (Race), Religion, Ethnicity, National Origin and Exercise ofConstitutional Rights

This category is the easiest one to determine. An individual will beconsidered “similarly situated” to another regardless of the fact that theyhave different religious beliefs or different countries of origin, and regardlessof the fact that one of them exercised a constitutional right while the otherdid not. The strict protection of the Constitution against discrimination onthe basis of these grounds is well established. A claim that two individu-als are not similarly situated because they differ in one of these charac-teristics is simply inconceivable to anybody who is familiar with Ameri-can constitutional law and history.89 With regard to prosecution, the Su-preme Court and States’ courts have stated many times that selectivity,based upon one of the above criteria, is unjustiªable and impermissible.90

Therefore, it is clear that, in our context, a distinction cannot be drawnbetween individuals along the lines of such “suspect” classiªcations. Forexample, a claim that a black defendant is not similarly situated to awhite person who committed the same crime and was not prosecuted,because the white is of a different religion or a different ethnicity, shouldbe unequivocally rejected.

2. Gender and Sexual Orientation

Two classiªcations that are no less arbitrary when it comes to selectiveprosecution than the previous categories are classiªcations based on gen-

89. The importance of protecting people who exercise fundamental constitutional rights

and “discrete and insular minorities” from discrimination was recognized in the fa-mous footnote, no. 4, of Justice Stone in United States v. Carolene Products Co., 304U.S. 144, 152 n.4 (1938).

90. See, e.g., Oyler v. Boles, 368 U.S. 448 (1962) (selectivity based upon “an unjustiªablestandard such as race, religion or other arbitrary classiªcation” is impermissible);People v. Serna 139 Cal. Rptr. 426 (1977) (The allegations that defendants were sin-gled out for prosecution solely because of their vigorous use of their First Amend-ment rights to protest the policies of the school and the school system, if proved,would constitute the defense of invidious prosecution); United States v. Smith 354A.2d 510 (D.C. 1976) (a policy intended to deter defendants from exercising their le-gal rights could not be tolerated). Murgia v. Municipal Court for Bakersªeld JudicialDist., 540 P2d 44 (Cal. 1975) (A selective enforcement policy directed against mem-bers or supporters of a labor organization is prima facie discriminatory and invalidunder the Equal Protection Clause).


der and sexual orientation. Yet, in general, the courts have not regardedthese classiªcations, and have not scrutinized them, under the same levelof suspicion that they have regarded classiªcations along the lines of race,ethnicity, or religion. I will brieºy claim that neither gender nor sexualorientation should ever be taken into account in the decision whether toprosecute or not. More important to the discussion here, I will argue thateven if one thinks that in some cases distinctions along these lines arepermissible, they should not be recognized as a distinction in our con-text—i.e., when a black defendant argues that a white who was not prose-cuted is similarly situated.

The Court regards gender classiªcations as less suspicious then racialor religious classiªcations. Gender classiªcations are not subject to the“strict scrutiny” standard of review that requires the government to showthat it is pursuing a “compelling interest.” On the other hand, genderclassiªcations are also not subject to the “rational relationship” test, ac-cording to which the court only checks if it is conceivable that the classiªca-tion bears a rational relationship to an end of government that is not con-stitutionally prohibited. The court adopted instead the “intermediate” test,according to which the gender classiªcation will be upheld only if thecourt ªnds it is “substantially related” to “an important government in-terest.”91

This difference in the general Equal Protection jurisprudence betweenthe protection granted to racial, religious, or ethnic groups and the pro-tection granted to women (or men) has permeated the ªeld of selectiveprosecution. Although the courts have usually stated that selectivitybased on gender is impermissible,92 they have not been as clear and un-ambiguous about it as they were in cases related to racial classiªcations.In fact, one cannot speak of “the courts” in this context, since they weredivided. Some courts even held explicitly that there is no constitutionalprohibition against selectivity based on sex.93

The under-protection from gender discrimination is subject to solidcritiques. It seems to me that the same reasons that justify strict scrutinyof race, religion, and national origin classiªcations—namely, the longhistory of stereotyping and subjugation of such groups—apply and shouldjustify heightened scrutiny of gender classiªcation. Even if we accept the

91. See generally, Rotunda & Nowak, supra note 87, at 216–26.92. See, e.g., People v. Superior Court of Alameda County, 562 P.2d 1315 (Cal. 1977) (held

that sex was an arbitrary classiªcation for the purposes of a discriminatory enforce-ment claim); State v. Maldonado 578 P.2d 296 (Mont. 1978) (stating that the consciousexercise of some selectivity in enforcement was not in itself a federal constitutionalviolation absent an allegation and a showing that the selection was deliberatelybased upon an unjustiªable standard or arbitrary classiªcation such as sex); State v.Evans, 326 S.E.2d 303 (N.C. Ct. App. 1985) (defendant failed to prove discriminatoryenforcement of the law based on sex, but the court regarded the claim, if proven, as avalid one). But see Boys Scouts of America v. Dale, 530 U.S. 640 (2000) (holding thatapplying New Jersey’s public accommodations law to require Boy Scouts to admitplaintiff, Assistant Scoutmaster who was expelled after he publicly declared he washomosexual, violated Boy Scouts’ First Amendment right of expressive association).

93. Minneapolis v. Buschette, 240 N.W.2d 500 (Minn. 1976) (held that, as a matter of con-stitutional interpretation, sex had not been declared an invidious or arbitrary basisfor discrimination).


view that “biological differences” in some instances justify different treat-ment of men and women in certain areas of the law, in the context of crimi-nal prosecution it is hard to think of reasons that will justify drawing adistinction on the basis of gender. Gender is no less arbitrary in this con-text than race, creed, or national origin.

The constitutional protection of gay people (or lack thereof) is evenmore complex. Formally, it is not a “suspect classiªcation,” nor is it “quasisuspect” as gender, although recent developments might suggest that thisapproach is likely to change.94 However, in my proposal this classiªcationis in the same category as religion, national origin, and gender because itis as arbitrary. In the current social context and in light of history, sexualorientation is susceptible to being a basis for discrimination, and classiªca-tions on such a basis justify strict scrutiny. In the context of criminal law,any distinction or selectivity drawn on the basis of sexual orientation isirrational.

The narrow question that has to be answered in this Article is whethera claim of a black defendant, in the ªrst stage of her selective prosecutionchallenge, that a white similarly situated to her was treated unequally, canbe dismissed on the basis that the white persons’ gender or sexual orien-tation was different? The answer to this question should be negative. Ifthere is no rationale for distinguishing between potential defendants onthe basis of their gender or sexual orientation, there is also no rationalefor accepting gender or sexual orientation as a distinction under the pro-posal of this Article.

B. Class

Let us assume man to be man, and his relations to the world to bea human one. Then love can only be exchanged for love, trust fortrust, etc. If you wish to enjoy art you must be [an] artistically cul-tivated person; if you wish to inºuence other people, you must bea person who really has a stimulating and encouraging effectupon others . . . . If you love evoking love in return, i.e., if you arenot able, by the manifestation of yourself as a loving person, tomake yourself a beloved person—then your love is impotent anda misfortune. —Karl Marx, 184495

Mostly, however, dominance is a more elaborate social creation,the work of many hands, mixing reality and symbol. Physical

94. Especially important is the Supreme Court decision in Romer v. Evans, 517 U.S. 620

(1996) (invalidating Colorado’s Amendment 2 that had forbidden sexual orientationanti-discrimination legislation on the grounds that it did not pass the rational basisreview). This decision is a sign of hope, after the infamous decision in Bowers v.Hardwick, 478 U.S. 186 (1986) (upholding a Georgia sodomy statute on the groundthat there is no fundamental right to private, consensual homosexual sodomy underthe Due Process Clause). See also Courtney G. Joslin, Recent Development: Equal Protec-tion and Anti-Gay Legislation: Dismantling the Legacy of Bowers v. Hardwick, 32 Harv.

C.R.-C.L. L. Rev. 225 (1997).95. Karl Marx, Economic and Philosophical Manuscripts, in Early Writings 193–94 (T. B.

Bottomore ed., 1963).


strength, familial reputation, religious or political ofªce, landedwealth, capital, technical knowledge: each of these, in differenthistorical periods, has been dominant; and each of them has beenmonopolized by some group of men and women. And then allgood things come to those who have the one best thing. Possessthat one, and the others come in train. Or, to change the metaphor,a dominant good is converted into another good, into many oth-ers, in accordance with what often appears to be natural processbut is in fact magical, a kind of social alchemy. —Michael Walzer, 198396

In Marx’s utopia, different social goods have their own “spheres” ofoperation. This is not the world in which we are living. We live in a worldin which, as in Walzer’s depiction, people dominate all spheres if theyhave the one social good. In our times, as both Walzer and Marx aguishlydescribed, this social good is money. Money controls all spheres. If youhave enough money, you will be able to enjoy the arts, you will havemore political inºuence, your chances to be “in love” are higher and youwill probably get more “justice.”

When it comes to law, we feel strongly that equal protection for poorand rich alike is an essential condition for a just system. In Walzerianterms, the sphere of law is ideally insulated and impenetrable to theinºuences of money, political positions, familial ties, and so forth. As Jus-tice Jackson put it:

We should say now, and in no uncertain terms, that a man’s mereproperty status, without more, cannot be used by a state to test,qualify, or limit his rights as a citizen of the United States. “Indi-gence” in itself is neither a source of rights nor a basis for denyingthem. The mere state of being without funds is a neutral fact—constitutionally an irrelevance, like race, creed, or color.97

But the Court did not stand behind this ideal vision, and, in fact, re-garded class, unlike race, creed or color, as relevant to people’s rights.Like gender classiªcations, class classiªcations have not been recognizedby the Court as “suspect” and therefore are not subject to strict scrutiny.Even the egalitarian-minded Warren Court did not recognize explicitlypoor people as a “suspect” group for the purposes of equal protectionclaims. As Professor Tribe describes it critically, the courts were willing toprovide “minimal protection” for the poor but not “equal protection.”98 Isubscribe to Tribe’s critique, which is extracted in the following Walzer-inspired paragraph:

Personal qualities and social goods have their own spheres of op-eration, which are governed by different principles of distribution:welfare to the needy, health care to the inªrm, honors to the de-serving, political inºuence to the persuasive, salvation to the pi-

96. Michael Walzer, Spheres of Justice: A Defense of Pluralism and Equality 11

(1983).97. Edwards v. California, 314 U.S. 160, 184–85 (1941).98. Tribe, supra note 49, at 1626–72.


ous, luxuries to those inclined and able to pay for them. Injusticemay result, however when the distribution principle of one sphere,such as material wealth, is allowed to invade the spheres of othersocial goods and determine who gets what. The end result may benot just an inequitable distribution of social goods, but the subju-gation of those people who do not possess that particular item bywhich all other social goods are valued.99

We would like to think that in criminal law, money cannot buy justice.But it seems that, apart from the obvious way money penetrates thecriminal justice system through the quality of defense that money can andactually does buy, money also exerts inºuence through selective law en-forcement policies. By focusing enforcement resources and efforts, espe-cially in regard to consensual crimes (like drug-related offences), on cer-tain neighborhoods, the poor are discriminated against. Money, thus,penetrates to the sphere of criminal justice in which the determination ofwhether people are arrested and punished should be according to theirobedience or disobedience to the law and not according to their wealth.

Drawing on recent legal scholarship, I identify three main argumentsthat can be used to support selectivity in law enforcement that is based onclass. The ªrst argument builds on the view expressed by Randall Ken-nedy regarding the controversy over the disparate impact of the crack-cocaine/powder-cocaine distinction.100 The powerful claim Kennedy makescontests the assumption of many critics of the crack-cocaine/powder-cocaine distinction, according to which the harsher punishments regu-larly imposed on black crack offenders are necessarily a burden uponblacks.101 In some cases, according to Kennedy’s view, the law enforce-ment in black communities should actually be viewed as a “subsidy” ratherthan discrimination, because it is advantageous to the law-abiding blackpeople who are the main sufferers from criminal activity within the blackneighborhoods. The application of this argument to class would be thecontention that focusing law enforcement on poor neighborhoods is actu-ally beneªcial for the poor.

The second and third arguments in support of a heightened level oflaw enforcement in poor neighborhoods are predominantly related toconsensual crimes, and they are presented and discussed in detail in a1998 article by William Stuntz.102 The second argument is that, from theperspective of retributive justice, a crime that occurs in a low-class neigh-borhood (e.g., selling drugs in street markets) is worse than the same

99. Id. at 1659.

100. Anti-Drug Abuse Act of 1986, Pub. L. No. 99-570, 100 Stat. 3207. This act providedharsh minimum sentences for drug trafªcking in general, but it also distinguishedbetween crack-cocaine and powder-cocaine. The prescribed sentence for crack-cocaine trafªcking was set to be dramatically harsher. It is uncontroversial that themuch harsher penalties for trafªcking of crack-cocaine, as opposed to powder-cocaine, are imposed almost exclusively on black defendants.

101. Randall Kennedy, The State, Criminal Law, and Racial Discrimination: A Comment, 107Harv. L. Rev. 1255 (1994). See also Kate Stith, The Government Interest in Criminal Law:Whose Interest Is It, Anyway?, in Public Values in Constitutional Law 137, 153

(Stephan E. Gottlieb, ed., 1993).102. William J. Stuntz, supra note 33.


crime that occurs in upscale neighborhoods. Street crimes are often con-nected with violence and other crimes, they cause fear among the popu-lation, and they create disorder that in turn causes, at least according tothe “broken windows” theory,103 more criminality. If a crime committed bya poor person in a poor neighborhood has worse implications than thesame crime committed by a rich person in an upper class area, so the ar-gument goes, it is justiªable for law enforcement agencies to target pooroffenders rather than wealthy ones.

The third argument is about the cost of policing. Simply put, it seemsto be much easier, cheaper, and productive for the police to look for drugsin poor neighborhoods where people sell them on the streets. In a worldof limited resources, is it not legitimate to use the resources in a morecost-effective way? The logic of this type of argument can be taken fur-ther, to a point Stuntz has not mentioned, but it is of particular relevanceto the question of selective prosecution. It is probable that in many cases,the prosecution of a rich offender will be much more expensive than that ofa poor one, mainly because of the ability of the rich to put on a defensethat will be far more expensive for the prosecution to cope with. Is it, forexample, illegitimate for the prosecution, under these assumptions, tomake a decision to use its limited resources to prosecute ten poor defen-dants instead of prosecuting one rich defendant?

These are powerful arguments, but there are counter-arguments thatare no less convincing. I ªnd the argument that law enforcement in poorneighborhoods is actually good for the poor very hard to agree or dis-agree with. It is difªcult to tell what “the poor population” favors. Likeany other attempt to speak on behalf of a group, speaking for the poor is arisky task. There are many voices within any group and one should becareful not to generalize and thus silence some of the voices. However, itis important to note that Kennedy’s view in regard to black communitieswas an exception. Other scholars who wrote about this issue opposed thepolicies that Kennedy supported, and rejected his arguments.104 Partiallyborrowing from the critiques of Kennedy’s work, I will offer several cri-tiques of the possible contention that unequal treatment by the law en-forcement agencies based on class is actually good for the law-abidingpoor. First, the dichotomy of “criminals” and “law-abiding people” isproblematic on several levels.105 Second, it is not clear whether such une-

103. According to the Broken Windows Theory, crime can be reduced by addressing visi-

ble signs of community disorder. See James Q. Wilson & George L. Kelling, BrokenWindows, Atlantic Monthly, Mar. 1982, at 31 (“If a window in a building is brokenand is left unrepaired, all the rest of the windows will soon be broken . . . . One unre-paired broken window is a signal that no one cares, and so breaking more windowscosts nothing . . . . We suggest that ‘untended’ behavior also leads to the breakdownof community controls.”). For a brilliant critique of this theory and the policing poli-cies emanating from it, see Bernard E. Harcourt, Illusions of Order (2001).

104. Janai S. Nelson, Disparate Effects in the Criminal Justice System: a Response to RandallKennedy’s Comment and Its Legacy, 14 Nat’l Black L.J. 222 (1997); David Cole, TheParadox of Race and Crime: A comment on Randall Kennedy’s ‘Politics of Distinction,’ 83Geo. L.J. 2547 (1995); Skalansky, supra note 28; Dorothy E. Roberts, Race Vagueness,and the Social Meaning of Order Maintenance Policing, J. Crim. L. & Criminology 775.

105. (a) Charles J. Ogletree Jr., The Burdens and Beneªts of Race in America, 25 Hastings

Const. L.Q. 219, 245 (1998) pointed out that there is no such sharp disconnection


qual law enforcement policies indeed reduce the level of criminality. Third,increased law enforcement carries with it increasing opportunities forabuse of power from which the whole population suffers. Fourth, sendinga large number of a community’s members to prison creates other difª-culties within those communities. Fifth, the above argument ignores andexacerbates other class inequalities that exist anyhow in the criminal jus-tice system (like the quality of representation). Sixth, it disregards otherbetter and less violent ways to try to solve the crime problem in poorneighborhoods (education, housing, rehabilitative programs, etc.).

The argument that a crime committed in a lower-class area is worsefrom a retributive perspective is challenged by two counter-arguments. First,the additional harm caused by the fact that the crime is committed in alower-scale market usually constitutes separate punishable offences. Sec-ond and more important, normally the poor do not choose to be poor andthey are unable to elect committing the crimes in a place where the crimewill cause less harm than in their own neighborhoods.

Finally, the argument that law enforcement in a poor neighborhood ischeaper raises the question “cheaper for whom?” It might be cheaper forthe law enforcement agencies, but it carries with it a host of costs to thepopulation.

In addition to the above-mentioned problems, class classiªcations areharmful because of the way class and race are intertwined. In the UnitedStates, there is a strong correlation between poverty and racial minorities.This is especially true with regard to poverty in urban neighborhoods.106

As Stuntz says:

[W]e might easily have seen the same law enforcement patternswe have seen over the past dozen years in a society where racialdivisions did not exist. But in a society where racial divisions areall too real, decisions that have no racial cause may still have avery powerful racial meaning.107

Prohibiting racial classiªcations in criminal law enforcement, but al-lowing class consideration to be taken into account, is like kicking racediscrimination out through the main door but inviting it to reenter throughthe window.

between law abiders and lawbreakers. Many people are both victims of crime andvictims of law enforcement; (b) The distinction ignores the constitutive effect ofcriminal law. By adopting certain laws, policies, and punishment practices we deªneand construct people as criminals. This is Bernard Harcourt’s critic of the BrokenWindows theory, New York style policing, and other policies targeting the “disor-derly” See Harcourt, supra note 103, at 127–84; (c) The problem is not only with thedistinction itself, but also with allying with the law abiders against the interest of thelawbreakers who are often the weakest part of the community. Sheri Lynn Johnsonlaunched a similar critique on Kennedy’s “politics of respectability.” See Sheri LynnJohnson Respectability, Race Neutrality, and Truth, 107 Yale L.J. 2619 (1998).

106. Stuntz, supra note 33, at 1810; John D. Kasarda, Inner-City Concentrated Poverty andNeighborhood Distress: 1970 to 1990, 4 Housing Pol’y Debate 253, 263 (1993) (thepopulation of extremely poor neighborhoods in America’s 100 largest cities consistedof 57% blacks, 24% Hispanic, and 16% white).

107. Stuntz, supra note 33, at 1825.


I have been arguing that class classiªcations are bad. Now it is time toreturn to the key question of this Part of this Article: when a black defen-dant argues that a white similarly situated to her was not prosecuted, cana difference in their class or wealth negate the claim that they are similarlysituated? My assertion is that, even if one maintains that in some cases itis permitted for law enforcement agencies to distinguish between individu-als along class lines, the above question should be answered negatively.

The context in which a black defendant shows a white, allegedly simi-larly situated to her, who was not prosecuted, is already a suspect situa-tion. Allowing prosecutors to distinguish between the two on the basis thatone is rich and the other is poor increases the suspicion that such an ap-proach is nothing more than a disguise for race discrimination. Even ifthere is no intentional racial discrimination, the racially unequal treat-ment in which this kind of distinction might result is a matter of concern.

This is not to say that according to my proposal the wealth of the de-fendant should never be taken into account. It is just to suggest that itshould not be considered as a difference between allegedly black andwhite similarly situated individuals. Such a distinction, in other words,should not exempt the prosecution from moving to the second stage ofthis proposal—the presentation of statistical evidence.

The following hypothetical situation might help clarify my position. Aprosecution ofªce has a limited amount of money. There are eleven po-tential defendants accused of the same crime, say crack trafªcking. One ofthe potential defendants is a rich white person and the other ten are poorblack people. The prosecution knows that prosecuting the white will cost$10,000 because of her ability to pay for a high quality defense, and thatprosecuting each of the black suspects will have an average cost of $1,000.Suppose that the budget for crack-trafªcking prosecution at this time is$10,000. The prosecution makes the calculation and decides to not prose-cute the white suspect and thus to be able to prosecute ten other (black)offenders. The black defendants raise the claim of racially based selectiveprosecution. As part of the ªrst stage of proving their claim, they point tothe white suspect who was not prosecuted, and claim that she is similarlysituated to them. In this case, the fact that the white is rich and the blackdefendants are poor, should not, according to my proposal, negate theclaim for similarity. The state will have to present the whole statisticalpicture of the numbers of blacks and whites who were and were notprosecuted for similar conduct. If the statistics demonstrate that there isno general correlation between race and the decision to prosecute, theblack defendants will lose their claim. If the statistics show a correlationbetween race and the decision to prosecute for this kind of behavior, theblack defendants will win their selective prosecution claim. Thus, even ifone agrees that, in some cases, the prosecution can take into account classclassiªcations (a position that I, for the reasons stated above, tend to dis-agree with), these classiªcations will be prohibited if, in a particular case,the distinction also carries racial meaning, and the general statistics showthat blacks are treated unequally. This proposal is responsive to the dan-ger of letting race discrimination “reenter through the window,” as wellas to the need to amend situations of de facto unequal treatment.


C. Other Personal Circumstances

The fact that no person is identically situated to another raises thequestion of where to draw the line of distinction. A set of questions arisesin connection to personal circumstances other than those already men-tioned. Is it legitimate for law enforcement agencies to distinguish be-tween individuals on the basis of their familial status (e.g., married-single, parent-not parent)? Is it legitimate to make such distinctions on thebasis of medical condition (e.g., not to prosecute someone who is termi-nally ill)? What about distinctions on the grounds of people’s past (e.g.,criminal record, arrest record or benevolent histories like excellence incommunity service)?108 As opposed to the position I expressed regardingreligion, national origin, gender, and class, I do not think that all thesekinds of classiªcations are inherently bad. One could easily imagine cir-cumstances in which a decision whether to prosecute that takes into ac-count such consideration is a legitimate one. For example, the decisionnot to prosecute a terminally ill person and to prosecute a healthy personwho is otherwise similarly situated might be regarded as justiªable use ofprosecutorial discretion. The reason is that we usually do not think ofhealthy people as a group that is being discriminated against, and theclassiªcation seems not to be arbitrary but rather rational.

As one might expect, there is no clear ruling by the courts on thepermissibility of all such classiªcations based on “personal circum-stances.” In fact, there are very few cases that address this issue. Theprobable explanations for the lack of decisions are that it is taken forgranted that such classiªcations are within the scope of prosecutorialdiscretion, and that in many cases, nobody except the prosecution isaware of the fact that such distinctions have been made. In the severalcases in which the question arose, the courts were divided, but the gen-eral tendency was toward upholding the distinctions. Thus, for exam-ple, in some cases it was held that defendants’ occupations were not anunjustiªable basis for their selection for prosecution,109 while in other

108. Cooperation with the police and prosecution is another issue that might justify une-

qual treatment in certain circumstances. This issue raises myriad problems that arebeyond the scope of this Article. Most notable is the problem of reliability of snitchtestimony. See, e.g., Barry Scheck et al., Actual Innocence 126–57 (2000). For thepurpose of the problem presented in this Article, the arguments and conclusion inthe text regarding “personal circumstances” apply also to cooperation.

109. United States v. Wiley, 503 F.2d 106 (8th Cir. 1974) (The claim was that the govern-ment purposefully prosecuted attorneys for tax offences. Held that reasonable prose-cutorial discretion was permissible so long as the prosecutor did not discriminatebetween persons based upon an unjustiªable standard.); United States v. Swanson,509 F.2d 1205 (8th Cir. 1975) (Again, the claim was that a project giving special prior-ity to the prosecution of tax crimes by attorneys and certiªed public accountants wasnot unconstitutional discrimination. The court held that defendant failed to satisfythe showing of invidious prosecution, that is, prosecution based upon impermissibleconsideration). See also United States v. Kearney, 436 F. Supp. 1108 (S.D.N.Y. 1970)(held that defendant, a former FBI agent, had not shown that his selection for prose-cution was invidious, where he sought discovery of decisions by the Department ofJustice not to prosecute present agent of the FBI or the CIA).


cases the courts said that such distinctions constituted unconstitutionaldiscrimination.110

Public prominence of the defendant was another personal character-istic that was raised by defendants in their claim for selective prosecution.Generally, it was held that the defendants’ prominence did not constitutean invidious or arbitrary basis for selection of the defendant for prosecu-tion.111

Some distinctions based on personal circumstances are more rationalthan others. It depends extensively on the particular circumstances of aspeciªc case. I do not intend to make a general statement here regardingthe desirability of allowing the prosecution to make such distinctions. How-ever, I do believe that when it comes to the question of whether to recog-nize such circumstances as a difference under the proposed SSI standard,the answer—as in religion, national origin, gender, or class classiªca-tions—should be a negative one. That is, the response should be negativeeven if it is assumed that taking such considerations into account issometimes a legitimate exercise of prosecutorial discretion.

The rationale of this position is the risk of racial discrimination em-bedded in permitting such considerations to be taken into account. Therisk stems from three reasons. First, there are inªnite ways of distin-guishing between individuals. If there are racist elements within thecriminal justice system, these elements would be able to justify almostany racist decision by citing racially neutral differences between blackdefendants and white ones. Second, some personal circumstances classiªca-tions could be correlated with race. Not only class, but also criminal rec-ord, for example, is biased criteria in the sense that, although they areformulated in racially neutral phrases, they actually have a racial mean-ing.112 The third reason is a more elusive one. It is related to the concept ofunconscious racism, but my acknowledgment of it is related to my per-sonal experience as a public defender in Israel.

One shameful phenomenon that Israel shares with the United States isthe inferior situation of minorities in the criminal justice system. In Israel,

110. United States v. Robinson, 311 F. Supp. 1063 (DC Mo. 1969). See also People v. Kail,

150 Ill. App. 3d 75, 501 N.E.2d 979 (Ill. App. 4th Dist. 1986), appeal denied, 508 N.E.2d732 (Ill. 1987) and cert. denied, 484 U.S. 827 (Police department’s policy of enforcingordinances against prostitutes, but not against other persons, constituted discrimi-natory enforcement as applied to arrest of defendant for failure to equip her bicyclewith a bell).

111. United States v. Peskin, 527 F.2d 71 (7th Cir. 1975), cert. denied, 429 U.S. 818; UnitedStates v. Ojala, F.2d 940 (8th Cir. 1976); Commonwealth v. Beneªcial Finance Co. 360Mass. 188, 275 N.E.2d 33, 52 A.L.R.3d 1143 (1971), cert. denied, 407 U.S. 910; People v.Barnwell 143 Misc. 2d 922, 541 N.Y.S. 2d 664 (1989) (Prosecution of high school prin-cipal for criminal possession of controlled substance was not selective undercounty’s formal policy of considering defendant’s position in community as relevantfactor in plea negotiations where there is no constitutional right to plea bargain.)

112. In a study regarding the inºuence of factors such as citizenship, home ownership,marital status, substance abuse, employment status, criminal history, time in the area,and education, on bail decisions, it was found that “substantial differences betweenwhites, African-Americans and Hispanics/Latinos were evident; For many of thesefactors, differences were also evident for Asian-Americans” Report of the WorkingCommittee to the Second Circuit Task Force on Gender, Racial and Ethnic Fairness inthe Courts, 1997 Ann. Surv. Am. L. 117, 311–20 (1997).


the most troubling disparity in the criminal justice system is the ethnicone, namely, the disparity between Arabs and Jews.113 As a public de-fender, I represented many Arab and Jewish defendants in criminal trials.I realized that one explanation for the disparity could be the way “per-sonal circumstances” of Arabs and Jews are perceived by the differentactors in the criminal justice system. In general, I found it much easier toconvince a judge or a prosecutor that a Jewish defendant had “personalcircumstances” or a life story that justiªed mitigation or leniency. At times, Ifelt that Arab defendants were treated as faceless. They had no personalcircumstances, no biographies; they were all the same—simply criminals.The fact that they came from a broken family, suffered from drug addic-tion, or had mental or physical disability was not of much interest to any-body. The experience was not racism—at least not conscious racism. I feltthat people could not identify with and feel compassion towards the per-sonal complexities of “the Other.” It is much easier for a Jewish prosecu-tor or judge to empathize with a Jewish defendant whose life story wasoften closer to her own.

Similarly, in the American criminal justice system, white prosecutors114

can identify with, or at least can more easily understand, white defen-dants.115 The danger of classiªcations based upon personal circumstancesis that blacks’ personal circumstances will be invisible or at least less visi-ble than those of whites in the eyes of usually white prosecutors.

113. See Arye Rattner & Gideon Fishman, Justice for All?: Jews and Arabs in the

Israeli Criminal Justice System 45–61 (1998) (dealing speciªcally with the prose-cutorial decision to bring charges); Muhamad Salim Haj Yihye et al., The MagistrateCourt and its Functioning with the Minorities in Israel, 4 Pelilm 157 (1994) [Hebrew](ªnding, among other things, that the prosecution’s petitions to aggravate the pun-ishment caused an increase in the average punishment of Arabs and a widening ofthe gap between the average punishments of Jews and of Arabs); Yael Hassin, Mi-nority Juvenile Delinquents in Israel and the Social Response to Them, 17 Soc’y & Welfare

Q. for Soc. Work 283 (1997) [Hebrew] (ªnding that there has been a double stan-dard in the adjudication and punishment of minors in the 1980s and 1990s in Israel: asevere standard for Arabs and a more lenient one for Jews).

114. In a study regarding state prosecutors in death penalty states, it was found that “theprosecutors with ultimate charging discretion in death penalty states are almost en-tirely white. Of the 1838 total prosecutors in death penalty states, 1794 are white(97.5%), 22 are black (1.2%), and 22 are Hispanic (1.2%). In fact, in 18 of the 38 deathpenalty states, whites constituted 100% of the prosecutors. In contrast, 1538 of the3269 people on death row are white (47.1%), 1340 are black (41.0%), and 227 are La-tino (6.9%).” Jeffrey J. Pokorak, Probing the Capital Prosecutor’s Perspective: Race of theDiscretionary Actors, 83 Cornell L. Rev. 1811, 1818 (1998). In Orange County, Cali-fornia, which ranks high among the death penalty counties in the country, the deci-sion whether to seek the death penalty used to be taken by a panel of all white maleprosecutors. See Rene Lynch, Deciding on Life or Death for O.C.’s Worst Murderers; Jus-tice: Prosecutors meet in secret to weigh evidence, impact on jury. Critics liken process to‘Star Chamber.’, L.A. Times, Feb. 23, 1994, at A1.

115. Angela Davis’s description of a true case from the Public Defender Service for theDistrict of Columbia has much in common with the experiences I had at the PublicDefender Ofªce in Tel-Aviv. See Davis, supra note 7, at 38 (“The prosecutor wouldprobably deny that race or class had anything to do with the decisions made inMcKnight’s case. His unconscious racial biases, however, may have played a signiª-cant role in the process. It is doubtful that the white male prosecutor empathizedwith the middle-aged Vietnamese immigrant; it is likely that he would identify withthe defendant who was a white male college student.”).


To sum up, classiªcations based upon personal circumstances are notinherently bad. Sometimes they are legitimate and useful; however, onemust recognize that they could have racial meanings and consequences. Ido not propose to forbid such classiªcations, but when race is involved Ipropose stricter scrutiny. This will be achieved through not recognizingpersonal circumstances as a difference when a black defendant claimsthat a similarly situated white was not prosecuted. To reiterate, this doesnot mean that the prosecution will not be able to favor a white person be-cause of his personal circumstances. It only means that if the prosecutiondoes so, the burden will be shifted, and it will have to prove by statisticalevidence that, in general, there is no correlation between the decision toprosecute and the race of the potential defendant. Thus, if, for example,the prosecution uses normally a classiªcation that has a racial meaning, itwill probably be evidenced by the statistics and will be precluded. But ifthe classiªcation is racially neutral—not only on its face but also de facto—there is no reason to assume that the statistics will show bias. If the datashows that there is unequal treatment, it means that there is a serious ºawin the system. This ºaw justiªes the prohibition of prosecuting a blackdefendant while a white person who allegedly committed the same of-fence is not being prosecuted, even if there are differences in their per-sonal circumstances.

I am aware of the following difªculty that my proposal raises: it mightbe argued that although the proposal permits the prosecution to take intoaccount personal circumstances, it will create a disincentive to do so be-cause of the burden the prosecution will have in presenting statisticaldata under the second stage of the proposal. Thus, implementation of theproposal might entail situations in which distinctions that we consider asnot merely legitimate but also positive and desirable will in actuality beprecluded. Let us take terminal illness as an example. Almost everybodywould agree that it is sometimes justiªed not to prosecute a person who isabout to die from her illness, while prosecuting a healthy similarly situ-ated individual. The challenge to my proposal is the claim that it might bethat the prosecutor will decide to prosecute a terminally ill white personin order to avoid the burden of presenting statistical data.

I assert that it is not reasonable to assume that in many cases theprosecution will refrain from such benevolent, racially neutral distinc-tions. As I explained above, the burden to provide the statistics is not asheavy as it might initially appear. Once prosecution ofªces, as is probablyrequired under the new proposal, adopt a method of keeping records, thisburden will be even lighter. The costs of prosecuting the ill person, forexample, might be even higher than the cost of providing the necessarydata. Moreover, if the prosecution only considers racially neutral factors,such as the health conditions of the defendant, there is no reason to as-sume that the statistics will demonstrate unequal treatment. Therefore,the prosecution will not be deterred by the need to provide statistics.116

116. I submit that, on the margins, there might be some instances in which the prosecu-

tion will nevertheless decide not to make a distinction that could have been madeunder the current legal regime. If the prosecution decides to prosecute, the personalcircumstances will still inºuence the sentencing stage. Under the Federal SentencingGuidelines, for example, the courts have to consider among other things the criminal


D. Classiªcations Related to the Criminal Conduct

Thus far, I have argued that under the proposed SSI standard nothingshould be recognized as a difference. I will now address the only classiª-cation that will be acceptable for distinction between individuals in theªrst stage of the SSI standard. This is a classiªcation regarding the actualcriminal conduct in question. The distinctions can be made on the basis ofthe type of criminal activity performed by the individuals, the extent oftheir criminal activity, and the quantum of it. I will argue that, althoughthese distinctions are ostensibly obvious ones to be made by the prosecu-tion, we still have to limit them and to take precautionary measures pre-venting them from being (ab)used in a discriminatory way.

The most natural distinction is according to the type of criminal ac-tivity. Nobody would challenge the fact that, for instance, Americanprosecutors have the power to charge someone who committed a seriousfelony and not to prosecute a person who committed a misdemeanor.This is true not only if the criminal activities of the individuals constitutedifferent offenses, but also when it constitutes the same offence. Thus, forexample, it is undeniable that if there is a general law that prohibits thepossession of drugs without making a distinction between different drugs, itwill be legitimate for the prosecution in some cases to prosecute thosewho possessed a certain drug and not to prosecute those who possessedanother drug, as long as there is a rational explanation for such a practice.The courts have indeed upheld in many cases such distinctions based onthe type of criminal activity.117 The extent or the quantum of illegal activ-ity is also acceptable in many cases as a distinction between individuals,which justiªes unequal treatment. Thus, for example, no one will doubtthe decision of a prosecutor to dismiss the charges for a theft of a $5 bill,

history and other speciªc characteristics of the defendant. See 28 U.S.C.A. § 994.There may be several advantages to considering personal circumstances in the sen-tencing stage over leaving it to prosecutorial discretion. Most important, the discre-tion under the sentencing stage is much more limited and controlled than the prose-cutorial discretion. This is especially true under the rigid rules of the Federal Sen-tencing Guidelines and similar guidelines that are currently intact in the vast major-ity of the states. The clear rules that the guidelines prescribe and the fact that judgeshave to give reasons for any deviation from these rules reduce the danger of unequaltreatment. Conversely, prosecutorial discretion as we know it today is usually not re-stricted by guidelines and not documented. A prosecutor that does not have to ex-plain her decision is more prone to be inºuenced by unconscious racism. Anotheradvantage of the sentencing stage is that it allows for context speciªc decisions, be-cause of the variety of types of punishments that the judge can use. For example, ajudge might be able to take into consideration drug addiction and order an alterna-tive for incarceration such as probation with a condition of drug treatment.

117. See U.S. v. Cantu, 557 F. 2d 1173 (5th Cir. 1977); Wheaton v. Hagan, 435 F. Supp. 1134(M.D.N.C. 1977) (upholding the distinction made by law enforcement ofªcials in ar-resting possessors of marijuana while not arresting possessors of alcoholic beverages);Pet. of Breen, 237 F. Supp. 575 (S.D. Tex. 1964), aff’d on other grounds, 341 F.2d 96 (5thCir. 1965), cert. denied, 386 U.S. 926 (upholding selectivity in prosecution under theTexas habitual offender statute, the court noted that none of the other offenders wasserving time for crimes which had any relation to that of the petitioner); People v.Garner, 139 Cal. Rptr. 838 (2d Dist. 1977) (recognizing that a distinction drawn be-tween bookmakers and bettors is a distinction on the basis of the difference in thekind of criminal conduct that is neither suspect nor invidious).


but to prosecute someone who stole a million dollars.118 The same is trueregarding a decision to drop the charges against a person who committedone misdemeanor as opposed to a person who committed the same mis-demeanor ten times. Again, the courts indeed upheld these kinds of prose-cutorial decisions.119

Of course, not every distinction based on the type, extent, or quantumof criminality is rational, and therefore some distinctions will be constitu-tionally invalid.120 Nevertheless, the courts usually defer to prosecutorialdiscretion. The prosecution is bound by its role as an agent of the State toenforce the law. In a reality of limited resources, it is understandable thatthe prosecution exercises some selectivity based upon the criminal con-duct. But this selectivity cannot be discriminatory. I will now point outsome dangers of discrimination that lie in the careless exercise of selectiv-ity based upon factors related to the criminal conduct.

First, we have to be aware of the possibility that what the prosecutionor the courts see as differences in the type or the extent of the illegal ac-tivity is actually informed by stereotypes regarding different offenders.An explanatory example of this risk can be taken from the case of People v.Rodrigues.121 The defendants, two homosexuals, were seen sitting in aparked car at a freeway reststop “engaged in kissing, hugging, and sit-ting, alternately, on each other’s laps. On one occasion one man was ob-served touching the other’s thigh.” The two men were arrested and sub-sequently convicted for engaging in “lewd and dissolute conduct.” In re-

118. The distinction among “type,” “extent,” and “quantum” of illegal activity is some-

times not very clear, and can be easily deconstructed. For instance, it seems that if inthe above example we use a pack of cigarettes instead of the $5 bill, the distinctionbetween the cigarettes thief and the million-dollar thief will be categorized as a dif-ference in the “type” and not in the “extent” or in the “quantum.” However, I ªnd noreason to deviate here from the terminology that is common in the legal literature.See, e.g., Robert Kevin Allen, Selective Prosecution: A Viable Defence in Canada?, 34 Crim. L.

Q. 414 (1992); John S. Herbrand, Annotation, What Constitutes Such DiscriminatoryProsecution or Enforcement of Laws as to Provide Valid Defense in Federal Criminal Pro-ceedings, 45 A.L.R. Fed. 732 (1979); John S. Herbrand, Annotation, What ConstitutesSuch Discriminatory Prosecution or Enforcement of Laws as to Provide Valid Defense inState Criminal Proceedings, 95 A.L.R.3d 280 (1979).

119. See People v. Carter, 450 N.Y.S.2d 203 (2d Dept. 1982). In that case, two black menwere the only participants in a barroom brawl between blacks and whites and wereset to be prosecuted for assault in a county that, as a rule, did not prosecute any par-ticipant in such incidents. The court ultimately held that the defendants did not meetthe burden of showing illegal discrimination between persons similarly situated.Among other reasons, the court noted that the defendants, unlike other participants,committed more culpable conduct by their use of dangerous instruments. See alsoState v. Savoie, 320 A.2d 164 (1974), rev’d on other grounds, 341 A.2d 598 (N.J. 1975),(upholding the decision to initiate prosecution only against two ofªcials who re-ceived more than $100, while others were involved in the same kind of corruptionbut with lesser amounts of money); State v. Steurer, 306 N.E.2d 425 (Ohio 1973), cert.denied, 416 U.S. 940 (1974) (holding that the exercise of prosecutorial selectivity wasnot constitutionally impermissible since no unjustiªable standard was used in a casewhere 65 persons engaged in the sale of securities without a license, but only 14 de-fendants who had sold 10 or more securities were charged).

120. See, e.g., 227 Book Ctr., Inc. v. Codd, 381 F. Supp. 1111 (S.D.N.Y. 1974) (holding thatenforcement of obscenity laws against adult bookstores but not against cinemas con-stitutes a violation of the Equal Protection Clause).

121. 63 Cal. App. 3d Supp. 1, 133 Cal. Rptr. 765 (1976).


jecting the claim of selective prosecution, which was based on one of thearresting ofªcers’ statement that he probably would not have arrested amixed couple for the same conduct, the court said that “such conduct onthe part of the arresting ofªcers would not be based on an ‘unjustiªablestandard,’ for appellants’ conduct was ‘lewd and dissolute,’ as deªnedabove, where the mixed couple’s would not appear to be.”122 The court, inthis case, actually conºated its own homophobic stereotypes, which weremanifested in its evaluation of the defendants’ conduct.

Rodrigues is a rare case in which such conºation is apparent from justreading the case. The reason is the disgraceful fact that homophobic views—unlike racist or misogynic views—are a type of arbitrary offensive classiªca-tion that are not only pervasive in our society, but can also be expressedrelatively freely without shame or censure. But one can easily imaginehow the same mixture of stereotypes and the deªnition of a “conduct”could occur in a racial context. In such a case, it would most probably beunconscious or concealed. For example, it might be that carrying aweapon by a black person would be seen differently than the same con-duct performed by a white person. In a system in which race and nationalorigin were approved by the courts as legitimate considerations for suspi-cion,123 it is not hard to imagine how it can also be taken into account byprosecutors when determining that one conduct is more “serious” ordangerous than another. The possibility to charge people for a certain be-havior with a certain offence, or with an aggravated offence based on thesame behavior “with an intention to commit a crime,” increases that dan-ger. In these instances, one might conclude that the defendant intended tocommit a crime from circumstances other than the objective conduct. Thedanger is that race will unconsciously or consciously be one of these “cir-cumstances.”124

Differences in the type, extent, or quantum of legal activity, like differ-ences in personal circumstances, can be found in any comparison of anypair of cases. Just as no person is identical to another, also no activity isidentical to another. This means that the prosecution could attempt to justifyalmost any selectivity on the basis of different conduct. The danger is that, insome cases, the same differences might be regarded as differences that

122. Id. at 5, 767.123. See U.S. v. Weaver, 966 F.2d 391 (8th Cir. 1992) (approving the fact defendant was a

“roughly dressed young black male” as one of the factors that could create reason-able suspicion); see also U.S. v. Martinez-Fuerte, 428 U.S. 543 (1976) (“We further be-lieve that it is constitutional to refer motorists selectively to the secondary inspectionarea at the San Clemente checkpoint on the basis of criteria that would not sustain aroving-patrol stop. Thus, even if it be assumed that such referrals are made largelyon the basis of apparent Mexican ancestry we perceive no constitutional violation.”).

124. A striking example is McQuirter v. State, 36 Ala. App. 707, 63 So. 2d 388 (1953). Inthis case the defendant, a black man, was charged and convicted with “an attempt tocommit an assault with intent to rape.” The only evidence, except for the complain-ant’s testimony that the defendant followed her, was the testimony of a police ofªcerregarding statements that the defendant made to him. The defendant denied both thefollowing and the alleged statements. The court said that “In determining the ques-tion of intention the jury may consider social conditions and customs founded uponracial differences, such as that the prosecutrix was a white woman and the defendantwas a Negro man.”


“make a difference” and in other instances they might be regarded as dif-ferences that do not make a difference.

As demonstrated above, in most cases when the defendants raised theclaim of selective prosecution, the courts approved the selectivity basedon differences related to the criminal activity. But it is interesting to seethe court’s reaction on one occasion in which the prosecution was the partywho claimed similarity between the defendant and other individuals.This unusual situation arose in State v. Holloway.125 The defendant thereargued that he was singled out for prosecution for tax offences because ofhis racial and political status as a black representative in the General As-sembly of the State of Delaware. The State cited the prosecution of ªveother individuals on the same charges. Defendant attempted to distin-guish these prosecutions from his own on the basis that the indictmentstherein contained Title 11 criminal charges as well as allegations of Title 30tax violations. The trial court rejected the distinction and pointed out thatdefendant’s investigation also started as a Title 11 criminal investigationthat did not culminate in an indictment. The court of appeals added that:

Furthermore, this Court is not in the business of determining theprosecutorial merit of any criminal conduct under investigationby the Attorney General; such is the exclusive province of thatOfªce alone. The election not to proceed against defendant onother possible concomitant offenses does not impact the constitu-tionality of the State’s prosecution of defendant on the instantcharges. To hold otherwise would impose an additional constrainton the exercise of the State’s discretion neither anticipated norauthorized by Yick Wo v. Hopkins, supra, and its progeny.126

My claim is that this passage reºects an attitude of exaggerated deferenceto the prosecution from which the courts should refrain. On its face, itseems that there was a signiªcant difference between the defendant andthe other individuals. In other cases, when the defendants claimed similar-ity to other individuals, as we have seen before, the courts did bother tocheck if there was a justiªable distinction, and in most cases, they foundthat there was. But this time, the court was “not in the business” of get-ting into the depth of the defendant’s claim.

What are the conclusions that should be derived from the above analy-sis? It is clear that in the American system, prosecutors can legitimatelyconsider aspects that are related to the criminal activity when decidingwhether to prosecute or not. But does it mean that differences in the type,extent, or quantum of criminality will be regarded as differences when ablack defendant points to a white who was not prosecuted under the SSIstandard? When asking the same question in regard to personal circum-stances, I reached the conclusion that, although such circumstances aresometimes legitimate for the prosecution to take into account, they shouldnot be accepted as a difference under the ªrst stage of the SSI standard. Inregard to differences related to the criminal behavior, it is impossible tomaintain the same position. Otherwise, everybody is similarly situated to

125. Supra note 67.126. Id. at 979.


everybody else, and in every case a black defendant would be able to showa white person who was not prosecuted—no matter what this white per-son did—and, thus, compel the prosecution to provide statistics regardingthose who were and those who were not prosecuted for any type of of-fence or criminal behavior. This is, of course, both irrational and impracti-cal.

The way to cope with the dangers I have described above is limited.The court will have to determine which “differences really make a differ-ence.” In a Common Law case-by-case method, the courts will decide onquestions like: Is offence A similar to offence B? Is offence X plus offenceY similar to offence X alone? Is offence X preformed in a way C that issimilar to offence X performed in way D? These kinds of questions do nothave one “correct” answer, but the important thing here is consistency,not correctness. The courts will have to use a lot of common sense anddecide which differences justify unequal treatment. Take, for example, thefollowing question: is possession of one kilogram of a certain drug similarto possession of one gram of the same drug, for the purposes of the deci-sion whether to prosecute? If we think that the answer to this question iseasy, one can further ask if one kilo is similar to half a kilo, and so on. Anypoint that will be chosen for drawing the line of distinction will be some-what arbitrary. But still a line must be drawn. This will be the only way toreduce the dangers described above. After a substantial body of law hasdeveloped, the courts will be able to enforce the equal application of thestandards that will be developed.

E. Interim Conclusion

Under the proposed SSI standard, a black defendant will have toprove, at the ªrst stage, that a white individual was involved in a similarcriminal activity as hers, and was treated differently. The prosecution,which probably has more access to relevant information, will be able topoint out differences that are related to the criminal activity of the indi-viduals and that justify the unequal treatment. The court will then decidewhether the differences justify the unequal treatment. The decision of thecourts will conform to precedents that will be developed, and will ensurethat the prosecution is complying with a set of standards of distinction.Differences that are related to arbitrary classiªcations such as differencesbased on religion, national origin, gender, or sexual orientation will not berecognized as differences for this purpose, nor will class differences ordifferences in personal circumstances.

If the ªrst stage is concluded in a successful showing by the black de-fendant that a white SSI was not prosecuted, and the prosecution is un-able to show that this white person is not similarly situated, the burdenwill shift to the prosecution to show by statistical evidence that there is nocorrelation between the decision to prosecute and the race of the defen-dant, as described in Part IV. If the prosecution is not able to make such ashowing, the defendant will win her selective prosecution claim and thecharges against her will be dismissed.


VI. Is the SSI Standard Impossible To Prove?

In reference to the standard set forth in Armstrong for discovery, Jus-tice Rehnquist wrote for the court:

The similarly situated requirement does not make a selective-prosecution claim impossible to prove. Twenty years before Ah Sin,we invalidated an ordinance, also adopted by San Francisco, thatprohibited the operation of laundries in wooden buildings. YickWo, 118 U.S., at 374, 6 S.Ct., at 1073. The plaintiff in error success-fully demonstrated that the ordinance was applied against Chi-nese nationals but not against other laundry-shop operators. Theauthorities had denied the applications of 200 Chinese subjects forpermits to operate shops in wooden buildings, but granted theapplications of 80 individuals who were not Chinese subjects tooperate laundries in wooden buildings under similar conditions.127

And Justice Breyer, concurring, said:

Were the “selective prosecution” defense valid in this case . . . , itshould have been fairly easy for the defendants to ªnd, not onlyinstances in which the Federal Government prosecuted African-Americans, but also some instances in which the Federal Govern-ment did not prosecute similarly situated Caucasians.128

On the other hand, many critics of Armstrong argued that the similarlysituated standard would be impossible to prove. Angela Davis, for exam-ple, wrote:

Armstrong leaves the ordinary criminal defendant with little hopethat he might ever prevail on a race-based selective prosecutionclaim and even less guidance on how he might do so. Even if theCourt had more precisely explained the quality and quantity ofevidence necessary to cross the discovery threshold, it is doubtfulthat most criminal defendants would be able to meet that stan-dard.129

The truth is probably to be found somewhere between these two po-sitions. It might not be “fairly easy” for a defendant to meet the burden,but it is deªnitely not an impossible task. Davis’s two main arguments insupport of her view are not convincing, or at least are not applicable ascritiques to my proposal. Her ªrst argument is that “the nature of racismand discrimination has changed signiªcantly since 1886 [the year Yick Wowas decided], racism and discrimination are not always overtly displayedor even intentional.”130 This statement is true, but it is more adequate as a

127. Supra note 5, at 466. Although I subscribe to Justice Rehnquist’s view that the stan-

dard is not impossible to prove, I think that his reliance on the only case in which theSupreme Court approved a selective prosecution defense is somewhat ironic. Thereason that Yick Wo succeeded in his claim was that it was not demanded that heprove discriminatory intent, a requirement that the Court demands today, and thatwas repeated in so many words by Rehnquist himself in Armstrong.

128. Id. at 476.129. Davis, supra note 7, at 47.130. Id. at 44.


critique of the discriminatory intent prong than of the similarly situatedindividuals requirement. Since under my proposal intentions are irrele-vant, the fact that discrimination is not overtly displayed will not havemuch of an effect on the ability to prove the claim. The second main ar-gument that Davis puts forth is that “most criminal defendants are indi-gent and thus incapable of hiring lawyers and experts to conduct the typeof investigation and report apparently necessary to obtain discovery.”131

The problem of the indigence of most criminal defendants and the lack ofadequate representation is indeed one of the most serious problems in thecriminal justice system.132 But its applications are certainly not unique inthe context of the need to prove similarly situated individuals. It is muchmore serious a problem when it comes to defendant’s attempt to prove adefense of insanity, to challenge the DNA evidence of the prosecution, orto perform any other task that involves costly methods. Proving similarlysituated individuals will not regularly demand hiring experts, although itmight be necessary in some cases to conduct investigations. In some cases,indigent defendants who use the services of a Public Defender Ofªce, oranother institution that is in charge of the representation of indigent peo-ple, might actually proªt from the advantages in data collection of suchinstitutions and from their ability to engage in empirical research.133

131. Id. at 47.132. The literature about this problem is vast. See Charles J. Ogletree, Jr. & Yoav Sapir,

Keeping Gideon’s Promise through Public Defender Ofªces: The American and Israeli Expe-riences, 28 N.Y.U. Rev. L. & Soc. Change (forthcoming 2003); Note, Gideon’s PromiseUnfulªlled: The Need for Litigated Reform of Indigent Defense, 113 Harv. L. Rev. 2062(2000); Jeff Rosentzwieg, The Crisis in Indigent Defense: An Arkansas Commentary, 44 Ark.

L. Rev. (1991); Vivian O. Berger, The Supreme Court and Defense Counsel: Old Roads,New Paths—A Dead End?, 86 Colum. L. Rev. 9 (1986); Richard Klein, The EmperorGideon Has No Clothes: The Empty Promise of the Constitutional Right to Effective Coun-sel, 13 Hastings Const. L.Q. 625 (1986); William S, Geimer, A Decade of Strickland’sTin Horn: Doctrinal and Practical Undermining of the Right to Counsel, 4 Wm. & Mary

Bill Rts. J. 91 (1995). Stephen J. Schulhofer & David D. Friedman, Rethinking IndigentDefense: Promoting Effective Representation through Consumer Sovereignty and Freedom ofChoice for All Criminal Defendants, 31 Am. Crim. L. Rev. 73 (1993); Stephen B. Bright,Keeping Gideon from Being Blown Away—Prospective Challenges to Inadequate Represen-tation May Be Our Best Hope, 4 Crim. Just. 10 (1990) Bruce A Green, Lethal Fiction: TheMeaning of “Counsel” in the Sixth Amendment, 78 Iowa L. Rev. 433 (1993); Martin C.Calhoun, How to Thread the Needle: Toward a Checklist-Based Standard for Evaluating In-effective Assistance of Counsel Claims, 77 Geo. L.J. 413 (1988); Stephen B. Bright, Counselfor the Poor: The Death Sentence Not for the Worst Crime but for the Worst Lawyer, 103

Yale L.J. 1835 (1994); Welsh S. White, Effective Assistance of Counsel in Capital Cases:The Evolving Standard of Care, 1993 U. Ill. L. Rev. 323 (1993); Roscoe C. Howard, Jr.,The Defunding of the Post Conviction Defense Organizations as a Denial of the Right toCounsel, 98 W. Va. L. Rev. 863 (1996).

133. For the advantages of repeat players see the classic article, Marc Galanter, Why the“Haves” Come Out Ahead: Speculations on the Limits of Legal Change, 9 Law & Soc’y

Rev. 95, 98–100 (1974). On the potential of such institutions to collect data regardingracial disparity and the advantages of data collection in the death penalty context,see Carol Steiker & Jordan Steiker, Should Abolitionists Support “Reform” of the DeathPenalty? 63 Ohio State L.J. 417, 427–28. The public defender ofªce in Israel fre-quently uses its institutional ability to conduct empirical studies and provide statisti-cal data. In a case in which I was involved, we conducted a study on discriminationagainst Palestinians who were charged with illegal entry to Israel. See also the de-scription of Tuitt infra text accompanying note 135.


Two relatively recent cases, in which the defendants succeeded in trialcourts to prove the existence of similarly situated individuals who weretreated differently, and an order for discovery was granted, can give ussome idea of ways in which this requirement could be fulªlled. Fromreading the decision in United States v. Olvis,134 it seems that the defendantssimply knew of Caucasians who were involved in the same or relatedschemes, and were not prosecuted. In United States v. Tuitt,135 the defen-dants received information about similarly situated individuals from rec-ords of the Springªeld ofªce of the Committee for Public Counsel Serv-ices, which represents indigent defendants, and from the Hampden CountyBar Advocates Ofªce.

So, even under the current legal regime, it is not an insuperable taskto prove that similarly situated individuals from a different race weretreated differently. It will become easier under the modiªed SSI standardthat I am proposing. First, under the SSI standard there is only the need toªnd a single SSI while the current standard speaks of “individuals.” Sec-ond, under my proposal, “similarity” is deªned much more broadly. Third, Ipropose to change the rigid rules of evidence that apparently were ap-plied by the Court in Armstrong when it dismissed afªdavits presented bydefendants as “hearsay.”136 I do not intend to prescribe here the exact rulesof evidence that should apply, but I maintain that they can be more ºexiblethan the regular rules of evidence applicable at trial. The hearing is not infront of a jury, but in front of a judge, who can use her discretion to assessthe probative value of the evidence presented to her, instead of making abinary decision regarding its admissibility. Furthermore, the nature of theproceeding is that most of the information is accessible to the prosecutionand not to the defense. This in itself is a valid reason to allow the defenseto present evidence even if it is not admissible at the trial stage. If thatevidence contains inaccurate information, the prosecution will easily beable to expose that fact by using its own reliable information.

VII. Conclusion

Since Yick Wo v. Hopkins137 was decided 117 years ago, no defendant haswon a claim of selective prosecution in the Supreme Court. There were onlya handful of successful claims of racially based selective prosecution in anyother court, state or federal, although hundreds of defendants raise theissue every year.138 These facts speak for themselves. Either we can assume

134. Supra note 62. This case was reversed in Olvis, supra note 71, on the grounds that the

allegedly similarly situated individuals were actually not similarly situated. Withoutgoing into detail, it seems that under my suggestion for a revised SSI standard, theoutcome would have been the same as the one that the District Court reached.

135. Supra note 81.136. Supra note 7, at 47. Davis is noting that, although the court felt at ease in dismissing

sworn afªdavits by members of the bar presented by the defense as hearsay, thecourt had no problem accepting sworn afªdavits by police ofªcers. Id. at 46.

137. 118 U.S. 356 (1886).138. For a rare case in which the defendant ªnally prevailed in his racially based selective

prosecution claim, see People v. Ochoa 212 Cal. Rptr. 4 (1985) (dismissing chargesafter the prosecution refused to comply with discovery order on the issue).


that in the past 117 years there has not been any other case of racial discrimi-nation in prosecution, or the standard that the courts apply is wrong.

As formerly indicated, the SSI standard I am proposing is not in-tended to be a silver bullet that will solve the problem of racial discrimi-nation in the criminal justice system. Moreover, if adopted, it will surelyhave costs. The greatest cost involved is the statistical evidence that willhave to be provided, and the records that will have to be kept. Indeed, thecostly need to provide a comprehensive and not just a partial statisticalpicture in order for these statistics to represent an accurate reality wasprobably an incentive for the Court to stay away from such a standard,and to stick to the discriminatory intent requirement.

However, the potential advantages of the new standard, as describedin this Article, are greater than its costs. The proposed standard is aimedat identifying the phenomenon of unequal treatment and amendingsituations in which this phenomenon occurs. As I argued, racially basedselective prosecution is different in many respects from other equal pro-tection issues in different contexts. Its special features and the uniqueconcerns it raises make racially based selective prosecution a good placeto start releasing the equal protection jurisprudence from the captivity ofthe discriminatory intent requirement. This Article has suggested an al-ternative and has argued that it is a plausible one, while anticipating andresponding to foreseeable difªculties in its application.

Crime rates in the United States have decreased dramatically in thelast decade,139 but racial disparity in the criminal justice system continuesto increase.140 In 1960, African Americans comprised a third of the stateand federal prison population. In the early 1990s they already comprisedabout half of that population and the percentage is still on the rise.141 Ac-cording to a recent report by the Department of Justice, an estimated twelvepercent of African American men ages twenty to thirty-four are in jail orin prison.142 Amadou Diallo, Abner Luima, O.J. Simpson and Rodney King aremore than recent scandals. They are names that represent and underscorethe degree to which this country is divided between blacks and whiteswith regard to the criminal justice system. According to poll data, 72% ofthe black population believes that the criminal justice system treats blacksmore harshly than whites,143 and 66% of black citizens regard the criminaljustice system as racist.144 It is time to consider where we have gone wrong,and to propose reforms. This is the aim of this Article.

139. U.S. Dep’t of Justice, Bureau of Justice Statistics (Sept. 9, 2002), available at http://

www.ojp.usdoj.gov/bjs/glance/viort.htm.140. The empirical evidence is vast and devastating. For a good summary of some of the

data, see Cole, supra note 35, at 4–5.141. See generally Michael Tonry, Malign Neglect—Race, Crime, and Punishment in

America (1995).

142. U.S. Dep’t of Justice, Bureau of Justice Statistics (Apr. 2003), available at http://www.ojp.usdoj.gov/bjs/abstract/pjim02.htm.

143. See Stuntz, supra note 33 (citing The Gallup Organization, Special Reports: Black/White Relations in the U.S., at http://www.gallup.com/Special_Reports/black-white.htm (June 10, 1997)).

144. See Maria Puente, Poll: Blacks’ Conªdence in Police Plummets, USA Today, Mar. 21,1995, at 3A cited in Paul Butler, Racially Based Jury Nulliªcation: Black Power in theCriminal Justice System, 105 Yale L.J. 677, 699 (1995).

neither intent nor impact: a critique of the racially

Documents